Created attachment 29221 [details]
Xorg.0.log

Created attachment 29222 [details]
xorg.conf

In the log I have only added in the source of 2.6.0 the line in i830_sdvo.c
function i830_sdvo_dpms 
(i830_sdvo_dpms) to the message that says 
First %s output reported failure to sync 

in order to check that the message comes from there.

don't use zipped attachment, please.

Created attachment 29291 [details]
Xorg.0.log

as per Michael Fu request - non gzipped attachements (changed)

Created attachment 29292 [details]
xorg.conf

as per Michael Fu - xorg.conf not gzipped

Additional notes on the log I sent:

In the log the line:
(EE) intel(0): (i830_sdvo_dpms)First SDVOB output reported failure to sync

The "(i830_sdvo_dpms)" text is added by me in the driver to identify which part of the code yields the "...SDVOB output reported failure to sync"

Now this message is always there for the current shipment of the machines.(rev10)
On the previous ones (rev2) - the failure to sync appeared now and them at 100-200 X start-ups with the same behavior (black screen). Anyway - if not ideal - it was acceptable to have X not working one at 200 start-ups.

Whenever the "...SDVOB output reported failure to sync" did not appear in the log - the screen displayed properly.

Now it exhibits the behavior permanently at each start-up.

What I attempted so far with no succeess:
1) To increase the sdvox registers writing from 2 to 4 times - with no luck.
2) To add some small delay with usleep between each call to write sdvo registers 
3) Found on internet some proposal of rewrite of i830_sdvo_dpms that loops on checking that SDVO is re-enabled after each call. (inside the function implementation)

I am not able at the moment to identify the correct source of this synchro failure. 

Adrian

Hello
IMPORTANT - if you think you can make it faster we can ship a machine preinstalled with Opensuse 11.0 or 11.1 or Fedora or Ubuntu. These are os-es that we tried and they all do the same.
Then we will get it back by sendind a transport company to pick it up ?


Adrian

Important information:
I can see that on the new config after X launches but display does not sync issuing xrandr gives
Screen 0: minimum 320 x 200, current 1024 x 768, maximum 2048 x 1200
VGA connected 1024x768+0+0 (normal left inverted right x axis y axis) 0mm x 0mm
   1024x768       60.0*+
   800x600        60.3
   640x480        59.9
TMDS-1 connected 1024x768+0+0 (normal left inverted right x axis y axis) 376mm x 301mm
   1024x768       60.0*+   60.0
   1280x960       60.0
   1024x600       60.0
   800x600        60.3     60.0     56.2
   768x576        60.0
   640x480        60.0     59.9
TMDS-2 disconnected (normal left inverted right x axis y axis)

which means to outputs connected - but I don't have anything on the VGA port

The old version of the machines that are working xrandr yields:
Screen 0: minimum 320 x 200, current 800 x 600, maximum 2048 x 1200
VGA disconnected (normal left inverted right x axis y axis)
TMDS-1 connected 800x600+0+0 (normal left inverted right x axis y axis) 376mm x 301mm
   800x600        60.3*+   60.0     56.2
   640x480        60.0     59.9



So VGA is correctly detected as not connected. 

Will you please add the modedebug option in xorg.conf and attach the Xorg.0.log?
    Option "modedebug" "true"

Will you please also attach the vbios.dump? You can get the vbios.dump by using the following commands
    1. echo 1 > /sys/devices/pci0000:00/0000:00:02.0/rom
    2. cat /sys/devices/pci0000:00/0000:00:02.0/rom >vbios.dump
    3. echo 0 > /sys/devices/pci0000:00/0000:00:02.0/rom

Thanks.
    

the driver you use is too old - 2.6.0, would you be able to try a more recent liveCD , say Ubuntu 9.10 alpha?

Created attachment 29373 [details]
Xorg.0.log with "modedebug" "true"

Xorg.0.log - modedebug true

Created attachment 29374 [details]
video bios dump

video bios dump

Tried ubuntu 9.10 alpha -Same thing - no display - Xorg.0.log from ubuntu attached as well. Driver version in this case is 2.8.1

Meanwhile I opened two machines - one that is working and one not working. Altough there seem to be no differences as reported by lspci (other than revision 10 on chipset and VGA) the motherboards are a bit different -i.e. is the same board but redesigned - not all components are in the same places. 

Tested windows XP SP2 which works with an old driver 19/12/2007 intel.

What else ?
Question - for us the issue is blocking and vital - i.e. the bug priority has been changed to less important. 
How do you manage these priorities as for me not having anything displayed on the main screen is blocking and high priority ? It actually means that the machines are of no use. Using VESA is out of question as we use XV for displaying accelerated camera feeds.

Created attachment 29377 [details]
Xorg.0.log-ubunut-9.10-alpha

This is the Xorg.log.0 for ubuntu 9.10. In this case there is no xorg.conf. X is started with its internal quessed config.

so, how is the display configuration in the system? I mean, you mentioned all-in-one machine, so it has built-in display, right? You also said, There are two connectors at the back of the panel on VGA and one DVI, and both are free. the built-in monitor must have internal connection to the graphic device, right?

(In reply to comment #15)
> Created an attachment (id=29377) [details]
> Xorg.0.log-ubunut-9.10-alpha
> 
> This is the Xorg.log.0 for ubuntu 9.10. In this case there is no xorg.conf. X
> is started with its internal quessed config.
> 

pls try add i915.modeset=0 to kernel boot parameters and see if it works. there should be a default xorg.conf, please add Option "ModeDebug" "True" to the device section and attach xorg.log. thanks.

> pls try add i915.modeset=0 to kernel boot parameters and see if it works. there
> should be a default xorg.conf, please add Option "ModeDebug" "True" to the
> device section and attach xorg.log. thanks.

There is none that's for sure (xorg.conf). X is just started and it detects all things if there is a standard config (one display only,etc) This is the mode Fedora and Ubuntu are doing things since one or two versions of their distributions before. 
What I am going to do is to create a config file with X -configure then add in that one. 

Regarding your other question - display connectio 
Yes it sure must have an internal connection directly on the motherboard but things are quite squezed in there and I could not see it.
The computer is a cybernet Ione GX31 (19 inches) from 
http://www.cybernetman.com/default.cfm?DocId=8003

Strange however that when looking with xrandr it detects two connected outputs:
VGA and TMDS-1

Created attachment 29386 [details]
Xorg.log for i915.modeset=0 

Now it works for ubuntu 9.10 alpha with kernel 2.6.31

Created attachment 29387 [details]
xorg.conf generated with X -configure 

option modedebug true added

After adding i915.modeset=0 on the kernel command line for ubuntu 9.10 alpha

Ok so now I have a "visible" X with a black screen (not the usual gray dotter screen) - my first impression was is not working again. 

Not even the cursor is visible but if I start an xterm I can see its window.

How can we make it work for a stable release of a distribution - preferrably opensuse 11.0 or 11.1 ?

for 2.6.0 and 2.7.0 version of the driver they fail on 
SDVO_CMD_GET_TRAINED_INPUTS
They return 0 or false and that is where my message 
"First SDVOB reported failure to sync"
occurs.

I tried to add i915.modeset=0 on the command line of my kernel - 2.6.27 but is not recognized. I could add a custom kernel 2.6.31 to my system if that would solve the problem but I need to understand the problem.

Adrian

pls try these kernel patches, _without_ i915.modeset=0 set.

http://bugs.freedesktop.org/show_bug.cgi?id=22002#c25

( to see if VGA detection can be correct)


commit 942642a412454c3365f0abc8399c8ef2944f4eac
Author: Sean Young <sean@mess.org>
Date:   Thu Aug 6 17:35:50 2009 +0800


commit f8aed700c6ec46ddade6570004ce25332283b306
Author: Ma Ling <ling.ma@intel.com>
Date:   Mon Aug 24 13:50:24 2009 +0800


commit 7c8460db30dfd085ef3837c8fb02ecf2e718b983
Author: Zhenyu Wang <zhenyuw@linux.intel.com>
Date:   Tue Sep 8 14:52:25 2009 +0800

(above 3 are for SDVO multiplier and output clone mask, if you pull lastest linus's tree, these 3 have been integrated)

again, pls attach xorg.log with modedebug, and intel_reg_dump output. If you use Ubuntu, intel_reg_dumper is available in the xserver-xorg-video-intel-dbg package.

(In reply to comment #22)
> for 2.6.0 and 2.7.0 version of the driver they fail on 
> SDVO_CMD_GET_TRAINED_INPUTS
> They return 0 or false and that is where my message 
> "First SDVOB reported failure to sync"
> occurs.
> 

this proves another bug we've fixed in 2.8 driver about sdvo port mapping.

Michael,
 THere is nothing connected on the VGA output of the system if that's what you meant - To see if VGA detection is correct. If that's what your meant there is nothing to detect as there is no monitor attached on the VGA output. the correct one is the one it sees on TMDS-1.

Michael,
 THere is nothing connected on the VGA output of the system if that's what you
meant - To see if VGA detection is correct. If that's what your meant there is
nothing to detect as there is no monitor attached on the VGA output. the
correct one is the one it sees on TMDS-1.

(In reply to comment #23)
> pls try these kernel patches, _without_ i915.modeset=0 set.
> 
> http://bugs.freedesktop.org/show_bug.cgi?id=22002#c25
> 
> ( to see if VGA detection can be correct)
> 
> 
> commit 942642a412454c3365f0abc8399c8ef2944f4eac
> Author: Sean Young <sean@mess.org>
> Date:   Thu Aug 6 17:35:50 2009 +0800
> 
> 
> commit f8aed700c6ec46ddade6570004ce25332283b306
> Author: Ma Ling <ling.ma@intel.com>
> Date:   Mon Aug 24 13:50:24 2009 +0800
> 
> 
> commit 7c8460db30dfd085ef3837c8fb02ecf2e718b983
> Author: Zhenyu Wang <zhenyuw@linux.intel.com>
> Date:   Tue Sep 8 14:52:25 2009 +0800
> 
> (above 3 are for SDVO multiplier and output clone mask, if you pull lastest
> linus's tree, these 3 have been integrated)
> 
> again, pls attach xorg.log with modedebug, and intel_reg_dump output. If you
> use Ubuntu, intel_reg_dumper is available in the xserver-xorg-video-intel-dbg
> package.
> 

OK guys,
  I managed to get it working with Opensuse 11.1 + kernel 2.6.31 rc9 (and over) 

+ X.Org X Server 1.6.3.901 (1.6.4 RC 1) + intel driver 2.8.1.(repository download.opensuse.org/repositories/X11:/Xorg/openSUSE_11.1)

If you think guys you can make it work in a driver version that I can sleep in for xserver 1.5.2 that would be great as xserver 1.6 from that repository doesn't seem to be very stable. It crashes here and there - the packages don't have good dependencies and I am missing all updates regarding X this way ?

It doesn't work with Xserver 1.5.2 and intel 2.7.0 even with kernel 2.6.31

Up until 2.7.0 intel driver I can match it with xserver 1.5.2 - afterwards I need xserver 1.6. 

If something has been corrected it has after 2.7.0 version of the driver and only with support from kernel 2.6.31 or newer.

Can I hope for a backport to driver 2.7.0 of those corrections ? (maybe SDVO mapping is the problem) ?

Adrian

X -configure doesn't work !
I can only test for now 800x600. For the other resolutions I need to guess the pixel clocks ?
xrandr gives strange things (see some odd resolutions)
Screen 0: minimum 320 x 200, current 800 x 600, maximum 2048 x 2048
VGA connected 800x600+0+0 (normal left inverted right x axis y axis) 0mm x 0mm
   800x600        60.3*+   60.0     56.2
   640x480        59.9
TMDS-1 disconnected (normal left inverted right x axis y axis)
TMDS-2 connected 800x600+0+0 (normal left inverted right x axis y axis) 376mm x 301mm
   800x600        75.0*+   60.3
   1024x819       75.0
   1024x768       75.0     70.1     60.0
   800x640        75.0
   640x480        75.0     72.8     75.0     59.9
   720x400        70.1

The screen is capable of 1280x1024 for sure.

(In reply to comment #28)
> X -configure doesn't work !
> I can only test for now 800x600. For the other resolutions I need to guess the
> pixel clocks ?
Can you double check whether the following resolution can be used?
   1. 1024x819
   2. 800x640
> xrandr gives strange things (see some odd resolutions)
> Screen 0: minimum 320 x 200, current 800 x 600, maximum 2048 x 2048
> VGA connected 800x600+0+0 (normal left inverted right x axis y axis) 0mm x 0mm
>    800x600        60.3*+   60.0     56.2
>    640x480        59.9
> TMDS-1 disconnected (normal left inverted right x axis y axis)
> TMDS-2 connected 800x600+0+0 (normal left inverted right x axis y axis) 376mm x
> 301mm
>    800x600        75.0*+   60.3
>    1024x819       75.0
>    1024x768       75.0     70.1     60.0
>    800x640        75.0
>    640x480        75.0     72.8     75.0     59.9
>    720x400        70.1
> 
> The screen is capable of 1280x1024 for sure.
From the EDID it seems that the mode of 1280x1024 is reported. But it also reports that max pixel clock is 110MHz. 
So the 1280x1024@75Hz is marked as invalid as it is beyond the range of pixel clock.

If you can confirm that it can work on the mode of 1280x1024@75Hz, please attach the output of lspci -vxxx, dmidecode so that we can add it into the quirk table.
Thanks.
> 

(In reply to comment #27)
> OK guys,
>   I managed to get it working with Opensuse 11.1 + kernel 2.6.31 rc9 (and over) 
> 
> + X.Org X Server 1.6.3.901 (1.6.4 RC 1) + intel driver 2.8.1.(repository
> download.opensuse.org/repositories/X11:/Xorg/openSUSE_11.1)
> 
> If you think guys you can make it work in a driver version that I can sleep in
> for xserver 1.5.2 that would be great as xserver 1.6 from that repository
> doesn't seem to be very stable. It crashes here and there - the packages don't
> have good dependencies and I am missing all updates regarding X this way ?
> 
> It doesn't work with Xserver 1.5.2 and intel 2.7.0 even with kernel 2.6.31
> 
> Up until 2.7.0 intel driver I can match it with xserver 1.5.2 - afterwards I
> need xserver 1.6. 
> 
> If something has been corrected it has after 2.7.0 version of the driver and
> only with support from kernel 2.6.31 or newer.
> 
> Can I hope for a backport to driver 2.7.0 of those corrections ? (maybe SDVO
> mapping is the problem) ?
Now we have no plan to backport the SDVO mapping patch set to driver 2.7.0.
So it will be better to update into the latest intel graphics driver.
Thanks.
> 
> Adrian
> 

(In reply to comment #25)
> Michael,
>  THere is nothing connected on the VGA output of the system if that's what you
> meant - To see if VGA detection is correct. If that's what your meant there is
> nothing to detect as there is no monitor attached on the VGA output. the
> correct one is the one it sees on TMDS-1.
> 

what I mean is the driver should detect that there is _no_ monitor connected to VGA, but apparently the xrandr shows driver think there _is_ a monitor connected to VGA. That's why I aske you:
"
pls try these kernel patches, _without_ i915.modeset=0 set.

http://bugs.freedesktop.org/show_bug.cgi?id=22002#c25

( to see if VGA detection can be correct)
"
would you pls try the patch and see if it works ( no more VGA detected ) for you?

Michael,
  

> > 
> what I mean is the driver should detect that there is _no_ monitor connected to
> VGA, but apparently the xrandr shows driver think there _is_ a monitor
> connected to VGA. That's why I aske you:
> "
> pls try these kernel patches, _without_ i915.modeset=0 set.
> http://bugs.freedesktop.org/show_bug.cgi?id=22002#c25
> ( to see if VGA detection can be correct)
> "
> would you pls try the patch and see if it works ( no more VGA detected ) for
> you?

These seem to be driver patches and not kernel patches. Against which version of the driver do you want me to try them - and running on which kernel ?

Adrian

No,
   1024x819 is not supported - it is shown by the windows driver (intel) as well but when setting it from Intel Control panel it does nothing. When setting it from Windows - Display Properties it displays over the entire screen but there is a problem with the signal as the screen flickers alot and changes colors rapidly.
All the other resolutions are supported - on windows I get:
800x600
800x640
1024x768
1024x819 - not working ok
1152x864
1280x600
1280x720
1280x768
1280x960
1280x1024 - 75 Hz


On windows I have the choice for outputs on :
Monitor
Digital

It only works when dual display is activated.

If a activate Monitor I loose display. The same with Digital - so I think there is problem however with this hardware config - but windows manages to handle it better by default (you have display on the screen).

Adrian

TEST RESULTS


WINDOWS
On Windows if Monitor only or Digital only is activated as outputs from Intel's driver control panel - I totally loose display. Need to reboot in VGA mode +  uniinstall intel driver + re-install driver - I found no other way of doing it. 

LINUX
On linux it works only with intel driver 2.8.1 (2.7.0 doesn't work) and I finally been able to start it on a 2.6.27 kernel as well - without any 1915.modeset=0 parameter.  The only dependency required it seems to be the version of the driver. (kernel would be only for GEM and other stuff) 

The driver complains about DRM and memory manager (GEM) that are not available with 2.6.27 kernels but display is working 
Cannot change resolutions with xrandr - which is maybe ok given that I have an xorg.conf that specifies the resolution explicitely. 

I think there is a hardware issue with routing the video signal as well on this architecture (mothervoard + chipset). I think that the guys from 
http://www.cybernetman.com/ and especially those board designers in China are watching closely this thread as well in order to find a final proper solution. 

Adrian

LINUX
  Some EDID issue - driver thinks 1280x1024 is out of sync so this resolution cannot be used.
X -configure   crashes with 
X -configure                           
_XSERVTransSocketOpenCOTSServer: Unable to open socket for inet6
_XSERVTransOpen: transport open failed for inet6/os111:0        
_XSERVTransMakeAllCOTSServerListeners: failed to open listener for inet6

This is a pre-release version of the X server from The X.Org Foundation.
It is not supported in any way.                                         
Bugs may be filed in the bugzilla at http://bugs.freedesktop.org/.      
Select the "xorg" product for bugs you find in this release.            
Before reporting bugs in pre-release versions please check the          
latest version in the X.Org Foundation git repository.                  
See http://wiki.x.org/wiki/GitPage for git access instructions.         

X.Org X Server 1.6.3.901 (1.6.4 RC 1)
Release Date: 2009-8-25              
X Protocol Version 11, Revision 0    
Build Operating System: openSUSE SUSE LINUX
Current Operating System: Linux os111 2.6.27.29-0.1-pae #1 SMP 2009-08-15 17:53:59 +0200 i686
Build Date: 10 September 2009  01:41:57AM                                                    
                                                                                             
        Before reporting problems, check http://wiki.x.org                                   
        to make sure that you have the latest version.                                       
Markers: (--) probed, (**) from config file, (==) default setting,                           
        (++) from command line, (!!) notice, (II) informational,                             
        (WW) warning, (EE) error, (NI) not implemented, (??) unknown.                        
(==) Log file: "/var/log/Xorg.0.log", Time: Mon Sep 14 19:21:03 2009                         
List of video drivers:                                                                       
        unichrome                                                                            
        mach64                                                                               
        tseng                                                                                
        rendition                                                                            
        ati                                                                                  
        dummy                                                                                
        ztv                                                                                  
        cirrus                                                                               
        glint                                                                                
        vesa                                                                                                                                                               
        tdfx                                                                                                                                                               
        ark                                                                                                                                                                
        neomagic                                                                                                                                                           
        s3                                                                                                                                                                 
        chips                                                                                                                                                              
        radeon                                                                                                                                                             
        newport                                                                                                                                                            
        sis                                                                                                                                                                
        r128                                                                                                                                                               
        i740                                                                                                                                                               
        via
        mgahal
        v4l
        savage
        vmware
        sisusb
        geode
        trident
        i128
        radeonhd
        siliconmotion
        intel
        ast
        tga
        nsc
        apm
        s3virge
        mga
        voodoo
        nv
        fbdev

Backtrace:
0: X(xorg_backtrace+0x34) [0x8179b84]
1: X(xf86SigHandler+0xa1) [0x80edc22]
2: [0xffffe400]
3: /usr/lib/xorg/modules//drivers/via_drv.so [0xb7fd55d4]
4: X [0x80cd89f]
5: X(LoadModule+0x4f) [0x80cd9d1]
6: X(xf86LoadModules+0xf7) [0x80caab6]
7: X(DoConfigure+0xbc) [0x8104888]
8: X(InitOutput+0x1b9) [0x80c7f60]
9: X(main+0x27b) [0x80715d9]
10: /lib/libc.so.6(__libc_start_main+0xe5) [0xb7bb9705]
11: X [0x8071291]

Fatal server error:
Caught signal 11.  Server aborting


Please consult the The X.Org Foundation support
         at http://wiki.x.org
 for help.
Please also check the log file at "/var/log/Xorg.0.log" for additional information.

but my libSM does not matches the currect installed xorg-x11 packages cause it depends on a libuuid that I can't match (the opensuse installe with not let me force it against existing libuui) - but there are thing which are probably away alot of the driver problem,... and not regarding you guys.

Adrian

On Mon, Sep 14, 2009 at 10:23:43 -0700, bugzilla-daemon@freedesktop.org wrote:

> http://bugs.freedesktop.org/show_bug.cgi?id=23713
> 
> 
> 
> 
> 
> --- Comment #35 from asergiu <adrian@opticvideo.com>  2009-09-14 10:23:41 PST ---
> Backtrace:
> 0: X(xorg_backtrace+0x34) [0x8179b84]
> 1: X(xf86SigHandler+0xa1) [0x80edc22]
> 2: [0xffffe400]
> 3: /usr/lib/xorg/modules//drivers/via_drv.so [0xb7fd55d4]
> 4: X [0x80cd89f]
> 5: X(LoadModule+0x4f) [0x80cd9d1]
> 6: X(xf86LoadModules+0xf7) [0x80caab6]
> 7: X(DoConfigure+0xbc) [0x8104888]
> 8: X(InitOutput+0x1b9) [0x80c7f60]
> 9: X(main+0x27b) [0x80715d9]
> 10: /lib/libc.so.6(__libc_start_main+0xe5) [0xb7bb9705]
> 11: X [0x8071291]
> 
this one's a via driver bug, X -configure should work if you move it
away.

Guys,
   What else should I send you. For now I cannot use 1280x1024 resolution at all. 
I need to compile a 2.6.31 kernel for the latest version of the driver.
I need to compile Xorg with xserver 1.6.x 

The board manufacturer has been notified and confirmed there is an issue - which they haven't see with the Windows driver I guess, because the windows driver does display all resolutions - the only problem is when activating only one output when display goes black.

What do you need to make the driver perform correctly - even though maybe the hardware has some issues ? 

Adrian

(In reply to comment #37)
> Guys,
>    What else should I send you. For now I cannot use 1280x1024 resolution at
> all. 
> I need to compile a 2.6.31 kernel for the latest version of the driver.
> I need to compile Xorg with xserver 1.6.x 
As I said in comment #29 the pixel clock of 1280x1024@75HZ is beyond the max pixel limit, this mode is removed by xserver. 

Will you please try another monitor and see whether the higher resolution is provided?
Thanks.

> As I said in comment #29 the pixel clock of 1280x1024@75HZ is beyond the max
> pixel limit, this mode is removed by xserver. 
> 
> Will you please try another monitor and see whether the higher resolution is
> provided?
> Thanks.
> 

Unfortunately I don't have here right now a screen that does that resolution. I only checked on the same computer with Windows XP and the windows driver does not check for pixel clocks I guess  as the driver shows all possible resolutions of the video card (without taking in account the attached monitor) - so if you go to high - the monitor will go out of sync. 

XP - Up to 1280x1024- 75 all work except the one I mentioned 1024x819.

Cristau was right I removed the unichrome and radeon (as well) drivers as they are of no use to me.

Adrian

Created attachment 29653 [details] [review]
disable VGA plane before doing CRT hotplug detect

Another issue is still remaining. This is a "All in One" box, which means that there is no CRT monitor. But from the log it seems that the VGA is detected as connected.
Will you please try this debug patch and attach the Xorg.0.log? Please do the test in UMS mode. 

Thanks.

ping~ , asergiu...

Sorry Michael, 
I was away for a few days.

Is Ashley Lupton – in CC here from the part of the motherboard producer. He can put you in contact with his development division.

ashley.lupton@cyberneteurope.com

Adrian

We found that Opensuse release 11.2 M7 that has updated Intel graphics driver.
This installed successfully and gives video output to the LCD panel.
The default Intel driver is ver.2.8.1 in Opensuse 11.2 Attached log file provided for detailed analysis.

Please have ACEP download Opensuse 11.2 from below website to verify.
http://software.opensuse.org/developer

This new Linux OS has updated kernel, Xserver, and video driver to better support Intel graphics controller.


Please ask ACEP to confirm the above results and provide us with any feedback on this issue.

Created attachment 29921 [details]
ver.2.8.1 intel driver

(In reply to comment #43)
> We found that Opensuse release 11.2 M7 that has updated Intel graphics driver.
> This installed successfully and gives video output to the LCD panel.
I already managed to have the updated 
Xserver on opensuse 11.1 with the 2.8.1 intel driver. It does works on one display. 

Can you drive two screens with it ? 
1) Have you been able to drive a second display attached on the DVI-output or VGA and switch between LCD and DVI/VGA ?
2) What about mirroring the output on both LCD+DVI/VGA ?

Adrian

if you don't have any connection on the VGA port, there is still a problem - the VGA detection is wrong. is it possible for you to try the patch in https://bugs.freedesktop.org/show_bug.cgi?id=22002#c25 and upload your xorg.log?

Now it mostly works except that whenever the computers are cold (after a few hours being shutdown) - when X displays we get a sort of blueish screen. The things to display are there - only that there is some sort of very dense blue noise on the screen. Upon 5-10 minutes of poweron if I restart X the displays works correctly but not before that.

Right now I am using Xorg 1.6.5 with intel driver 2.8.1 from opensuse 11.1 X11: factory repository.

Here are some pictures of the behavior - problem and after a few minutes with X restarted and working correctly

http://opticvideo.com/intel/IMG_Blue1.jpg
http://opticvideo.com/intel/IMG_Blue2.jpg
http://opticvideo.com/intel/IMG_OK.jpg

X alone exhibits the same behavior - here is easier to see the issue.

Adrian

I have also installed opensuse 11.2 with a  2.6.31.5 kernel.
X.Org X Server 1.6.5
Release Date: 2009-10-11

Xserver - intel driver version 2.9.1.

I still get the blue screen. Is like an interference with screen (pictures from my messages above show the exact screen).

There seems to be no other solution.
Working now on applying the patch for disabling VGA plane detect - but I am really loosing hope here on this issue. 

A version of the Windows XP driver does the same thing - blue screen. 
we Have been provided with a new version now by the computer producer (cybernet - john birght is in copy of the messages here). They are stating that this new driver does not exhibit the blue screen behavior. 

What is strange is that on 50 computers bought not all show the sort of blue screen (with linux or windows)- some do it every time, others just now and then. 


Adrian

disable VGA plane before CRT hotplug is not giving anything new. I applied the patch to a 2.9.0 driver (latest stable from Intel LInux site) and I still get the VGA output detected as connected altough there is no VGA screen connected on the VGA screen.

The blueish screen might be a problem of screen frequencies missused because of the two outputs - both being detected as connected ????

Adrian

(In reply to comment #48)
> I have also installed opensuse 11.2 with a  2.6.31.5 kernel.
> X.Org X Server 1.6.5
> Release Date: 2009-10-11
> 
> Xserver - intel driver version 2.9.1.
> 
> I still get the blue screen. Is like an interference with screen (pictures from
> my messages above show the exact screen).
> 
> There seems to be no other solution.
> Working now on applying the patch for disabling VGA plane detect - but I am
> really loosing hope here on this issue. 
> 
> A version of the Windows XP driver does the same thing - blue screen. 
> we Have been provided with a new version now by the computer producer (cybernet
> - john birght is in copy of the messages here). They are stating that this new
> driver does not exhibit the blue screen behavior. 
> 
> What is strange is that on 50 computers bought not all show the sort of blue
> screen (with linux or windows)- some do it every time, others just now and
> then. 
> 
> 
> Adrian
> 

the blue screen issue more like some kind of loose of connection on color signals , rather than SW issue...

Created attachment 33033 [details]
Photo of the screen 1

Created attachment 33035 [details]
Photo of the display -2

I come back with the Intel card/monitor problem.
So:
xrandr sees:
os112:~ # xrandr
Screen 0: minimum 320 x 200, current 1024 x 768, maximum 2048 x 2048
VGA connected 1024x768+0+0 (normal left inverted right x axis y axis) 0mm x 0mm
   1024x768       60.0*
   800x600        60.3
   640x480        59.9
TMDS-1 disconnected (normal left inverted right x axis y axis)
TMDS-2 connected 1024x768+0+0 (normal left inverted right x axis y axis) 376mm x 301mm
   1024x819       75.0
   1024x768       75.0     70.1     60.0*
   800x640        75.0
   800x600        75.0     60.3
   640x480        75.0     72.8     75.0     59.9
   720x400        70.1


BUT THERE IS NO VGA display connected whatsoever. I just have an integrated LCD panel of 19 inches. 

I can make the display work in 800x600 by tweaking the vertical/horiz  refresh rate manually. But I am not succeding with other resolutions.

Also timings are not crrectly detected. 


This is a very old problem already. I am even ready to send a machine. I loaded openesuse 11.2 (kernel 2.6.31) which displays something now. Yet it is far from working. I still get the:
First SDVOC output reported failure to sync


Adrian

Sorry for the multiple postings.
On the last two attachements you can see that the screen is blue (like with some interference) - there is alot of blue noise on the panel.

It seems that with the latest intel windows driver it works ok

Adrian

OK I tried to apply the patch for VGA plane disable before CRT detect but it does not apply to intel 2.10.0 driver source tree.There is no i830_crt.c file and I don't know where to apply the patch in the new source tree.

I managed to advance a bit more meanwhile:

Compiling a 2.6.32.7 vanilla kernel with KMS on for i915 and disabling the VESA framebuffer (all framebuffers  in fact) I get a blueish display and xrandr reports this time VGA1 and DVI2 as connected (no VGA connected) - similar to before but :

If I issue:
xrandr --output VGA1 --off
(xrandr still reports VGA1 as connected afterwards - so the driver does not see there is no VGA)

xrandr --output DVI2 --auto 
it seems to go in 1024x768 (which is reported as maximum resolution for my LCD display) by xrandr. The LCD can do 1280x1024 - it does it on windows.

Guide me here guys as there is something wrong with this intel driver for Xorg.

Adrian

Created attachment 34137 [details] [review]
try the debug patch that uses the load-detection to check whether the VGA is connected

(In reply to comment #54)
> OK I tried to apply the patch for VGA plane disable before CRT detect but it
> does not apply to intel 2.10.0 driver source tree.There is no i830_crt.c file
> and I don't know where to apply the patch in the new source tree.

Sorry for the late response.

On this box the two issues are still remaining.
   a. The VGA is detected incorrectly.
   b. There is no display mode of 1280x1024@75Hz. The 1280x1024@75Hz is defined in the EDID. But as the pixel clock of 1280x1024@75Hz is beyond the max clock defined in EDID, it is removed. Similarly the 1280x960@75Hz is also removed. 

For the first issue: Will you please try the debug patch on 2.6.33 kernel and attach the output of dmesg ? Please add the boot option of "drm.debug=0x04".

thanks.

Created attachment 34618 [details]
Log of the DRM patch

I tried your patch. It commented out a sequence in intel_crt.c 
My kernel was vanilla 2.6.33.1 

Still the VGA looks connected after reboot. Here is the log 
for the drm modules loaded with debug=4

Tell me what else I can do.xrandr gives a strange resolution for the screen 1280x819. I have modesettings enabled in kernel and the X server version is :
 -version

X.Org X Server 1.8.0
Release Date: 2010-04-02
X Protocol Version 11, Revision 0
Build Operating System: openSUSE SUSE LINUX
Current Operating System: Linux os112 2.6.33.1-dadi #10 SMP PREEMPT Fri Apr 2 12:13:00 CEST 2010 i686
Kernel command line: root=/dev/sda2 resume=/dev/sda1 splash=silent quiet video=1280x1024@60
Build Date: 02 April 2010  10:34:59AM



xrandr yields
Screen 0: minimum 320 x 200, current 1280 x 1024, maximum 4096 x 4096
VGA1 connected 1280x1024+0+0 (normal left inverted right x axis y axis) 0mm x 0mm
   1280x1024      60.0*+   59.8
   1600x1200      60.0
   1400x1050      60.0
   1280x960       60.0     60.0
   1152x864       60.0
   1024x768       60.0     60.0
   800x600        60.3     60.0     56.2
   768x576        60.0
   848x480        60.0
   640x480        59.9     59.9     60.0
DVI1 disconnected (normal left inverted right x axis y axis)
DVI2 connected 1024x819+0+0 (normal left inverted right x axis y axis) 376mm x 301mm
   1024x819       75.0*
   1024x768       75.1     75.0     70.1     60.0
   800x640        75.0
   800x600        75.0     60.3
   640x480        75.0     72.8     60.0
   720x400        70.1

Ok I finally managed to hack the problem - but not the cause of it. 
I can force the display to show 800x600 and 1280x1024 without any blue screen issues. It only works by manually adding modelines for the 2 modes - forced at 60Hz (both 1280x1024 and 800x600).
The main cause is I think the fact that the driver sees the VGA output as connected (which is still the case) and tries to get display (EDID) information from it - or something related - which explains why I get the crap resolutions 

VGA connected 800x600+0+0 (normal left inverted right x axis y axis) 0mm x 0mm
   800x600        60.3*+   60.0     56.2
   640x480        59.9
TMDS-1 connected 800x600+0+0 (normal left inverted right x axis y axis) 376mm x
301mm
   800x600        85.1*+   72.2     75.0     75.0     60.3     56.2
   1280x1024      60.0
   1280x960       60.0
   1152x864       75.0
   1024x819       75.0
   1024x768       85.0     75.0     75.1     75.0     70.1     60.0
   832x624        74.6
   800x640        75.0
   640x480        85.0     75.0     72.8     72.8     75.0     75.0     60.0   
 59.9
   720x400        85.0     70.1
   640x400        85.1
   640x350        85.1



What I do is:
--------------
Delete all modelines from xorg.conf or with xrandr and leave only 
  Modeline "800x600_60.00"   38.25  800 832 912 1024  600 603 607 624 -hsync +vsync
  Modeline "1280x1024_60.00"  109.00  1280 1368 1496 1712  1024 1027 1034 1063 -hsync +vsync

Both work this time and the screen is not blue. 
The blue screen comes from 800x600 comes from the 75Hz I guess that is seen as capable on both screens VGA/TMDS-1. Only that TMDS-1 probably does not support it. 

This way I have sane 1280x1024 and 800x600 with 2 modelines created with cvt for 60Hz

Adrian

Does this still happen with current bits?  The kernel's EDID handling has had a few fixes recently.

AIUI, this bug has two elements:

1. The VGA remains connected after it is unplugged.

2. The modes for the TMDS are being misreported.

The workaround for (1) would be a xrandr --output VGA1 --off, and maybe with the change in VGA sense detection it may now be fixed. (2) may be fixed with the patches to fix EDID handling.

(1) sounds like it should be fixed with:

commit 149c36a346f63fa4be7d1432d7b1b3095a95bf47
Author: Adam Jackson <ajax@redhat.com>
Date:   Thu Apr 29 14:05:18 2010 -0400

    drm/i915: Be extra careful about A/D matching for multifunction SDVO
    
    If we're both RGB and TMDS capable, we'll have set up one connector for
    each.  When determining connectivity, require analog/digital state in
    the EDID block to match analog/digital support in the connector.
    Otherwise, both DVI and VGA will appear to be connected.
    
    Signed-off-by: Adam Jackson <ajax@redhat.com>
    Signed-off-by: Eric Anholt <eric@anholt.net>

I think this is working better now. The problem has been identified finally (match the capabilities of connectors so that we don't have 2 connected when is only one)

I do not have the hardware now - but if you guys need a test I can get hands on a piece like that.

Adrian

Please retest with latest kernels (at least v3.2) and report whether this is still an issue.

Created attachment 56882 [details]
Log files -dmesg & Xorg.log

Nope the problem is not solved.
I have loaded a 3.2.4 kernel from opensuse repos (and I've compiled one manually as well to be sure).

Here is what happens:
1) uname -a
Linux blue 3.2.4-1-desktop #1 SMP PREEMPT Sat Feb 4 14:19:04 UTC 2012 (484affb) i686 i686 i386 GNU/Linux

2) Kernel args:
kernel /boot/vmlinuz-3.2.4-1-desktop root=/dev/sda2 resume=/dev/disk/by-id/ata-ST3160318AS_9VM4S8PA-part1 splash=silent quiet showopts vga=0x314

CONCLUSIONS
a) There is no splash screen anymore due to KMS still not working- so that means that there is a problem starting there

b) X is not displaying anything afterwards 
xrandr shows:
blue:~ # xrandr 
Screen 0: minimum 320 x 200, current 1024 x 768, maximum 4096 x 4096
VGA1 connected 1024x768+0+0 (normal left inverted right x axis y axis) 0mm x 0mm
   1024x768       60.0* 
   800x600        60.3     56.2  
   848x480        60.0  
   640x480        59.9     59.9  
DVI1 connected 1024x768+0+0 (normal left inverted right x axis y axis) 376mm x 301mm
   1024x819       75.0  
   1024x768       75.1     75.0     70.1     60.0* 
   800x640        75.0  
   800x600        75.0     60.3  
   640x480        72.8     75.0     60.0  
   720x400        70.1  
HDMI1 disconnected (normal left inverted right x axis y axis)

Altough there is no VGA connection. I can see X output with vncviewer but not on the screen !
I do not see anything on the screen during boot (splashscreen) with KMS in the initrd. If I take out the KMS from initrd then only X gives me a black screen.

With nomodeset in the kernel cmdline I can get a working config by using intellegacy driver for X by turning off the VGA output and only using the other one (it is named TMDS-1 in this case)

If you need log files or even access to the machine just tell me. I am very keen as well in helping this out as I cannot properly use those machines right now.

Adrian

Two things:

- Please reboot into the new kernel with drm.debug=0xe attached to the commandline and grab the entire dmesg.

- Do you have 2 monitors connected to that machine or is the information for the DVI1 output bogus?

Created attachment 56895 [details]
dmesg with drm.debug=0xe

drm.debug=0xe

cat /proc/cmdline
root=/dev/sda2 resume=/dev/disk/by-id/ata-ST3160318AS_9VM4S8PA-part1 splash=silent quiet drm.debug=0xe vga=0x314

Dmesg attached 
KMS -enabled during boot (modules in initrd)

The computer is a ALL IN ONE PC - I would say that the correct connection is the DVI1 (in the case of KMS) and not the VGA one. Why ? When not using KMS at all and starting X with intellegacy driver in order to see something on the screen I need after launching it to disable both outputs and enable the TMDS-2 (in that case) -the VGA is off. 
I think the VGA should not be appeareing connected.

Adrian

Judging from the dmesg the kernel correctly detects a screen on DVI and can parse the modelines you're also getting when using nomodeset and the legacy usermodesetting driver. We still detect a spurious VGA input.

Now the xrandr output disagrees with what the kernel detects, so can you double-check whether your running a recent intel ddx and don't have anything in your xorg.conf that causes this? Also attach the Xorg.log when running on a new kernel, too. I've noticed that the xorg.conf you've posted contains modelines. If you still use that, please remove your xorg.conf.

You've mentioned that when manually add modelines to VGA1 that exactly match the modes from the digital panel and enable both output, things work. Is that still the case with the latest kernel?

Created attachment 56923 [details]
Xorglog and X11config

Attached is the Xorg.0.log when running with KMS and the X11 config files (xorg.conf.d directory). I have tested with and without modesettings but with KMS I do not seem to be able to get output (modessetings line in 50-monitor.conf do not seem to help this time). I will try more tomorow.

Intel driver version is 2.12 which is a bit behind latest stable version 2.17.

I could also try with the latest but that means half a day to get a x86 dev system to get the sdks and compile the intel driver.

P.S. The X conf is clean - just defaults (Screen, device and monitor)

Adrian

Created attachment 56970 [details]
Xorglogdmesg

Kernel 3.2.4 - intelr driver 2.16 (previous before Q4 2011 released driver)
Kernel args: root=/dev/disk/by-id/ata-ST380815AS_6QZ7936E-part3 resume=/dev/disk/by-id/ata-ST380815AS_6QZ7936E-part1 splash=silent quiet selinux=0 drm.debug=0xe vga=0x317

There is still a problem. There is no image at all when booting with KMS -so that means KMS is not working. 

There is no way see something after boot - but even booting in runlevel 3 woul ot show anything with KMS.

There is only one output that appears connected this time: DVI1 (digital output)

For me there is still a problem with the KMS - otherwise there should be something on the screen during boot. I think once this is solved the X part will likely wotk as well no ?

Adrian

I guess there is no point compiling 2.17 ver of xorg intel driver as long as with KMS I have a black screen during boot (and a black screen after)

Adrian

Timeout. Please do reopen if you can still reproduce the issue and help us diagnose the problem, thanks.