[g35] [hardy alpha 5] xserver autoconfigure fails ONLY over HDMI (Intel G35)

[TheKorn]

Binary package hint: gdm

monitor: hanns-g hc194d
graphics chipset: intel G35
motherboard: asus p5e-vm hdmi
OS version: hardy alpha 5, patched on 3/5/2008

DISTRIB_ID=Ubuntu
DISTRIB_RELEASE=8.04
DISTRIB_CODENAME=hardy
DISTRIB_DESCRIPTION="Ubuntu hardy (development branch)"

gdm:
  Installed: 2.20.3-0ubuntu6
  Candidate: 2.20.3-0ubuntu6
  Version table:
 *** 2.20.3-0ubuntu6 0
        500 http://us.archive.ubuntu.com hardy/main Packages
        100 /var/lib/dpkg/status

When monitor is connected via VGA D-sub cable, everything works great! When monitor is connected via DVI cable (w/ DVI-HDMI adapter), monitor shows "unsupported display mode".

FWIW, this exact combo works GREAT under gutsy (even under HDMI).

more info: I'm rebooting in between swapping cables, and at no point in time are both cables connected to the machine. Sequence is to power down, swap VGA for DVI cable (or vice-versa), power up, see BIOS and then ubuntu splash screens, then when the login screen should come up (I hear the bongo sound, and can log in blind or administer via SSH) that's when I get unsupported display mode. So I'm not crashing at all, it's just not exactly working quite right.

[lspci]
00:00.0 Host bridge [0600]: Intel Corporation 82G35 Express DRAM Controller [8086:2980] (rev 03)
     Subsystem: ASUSTeK Computer Inc. Unknown device [1043:8295]
00:02.0 VGA compatible controller [0300]: Intel Corporation 82G35 Express Integrated Graphics Controller [8086:2982] (rev 03) (prog-if 00 [VGA controller])
     Subsystem: ASUSTeK Computer Inc. Unknown device [1043:8276]

[Kevin DeKorte (kdekorte-gmail)]

Log files are with VGA output "ignored", but with output still going to the VGA output

[Kevin DeKorte (kdekorte-gmail)]

Created an attachment (id=13587)
xorg.conf

VGA output is ignored

[Kevin DeKorte (kdekorte-gmail)]

Created an attachment (id=13588)
Xorg Log file

VGA output is ignored, and the log shows this, but output is still going to the VGA cable

[Kevin DeKorte (kdekorte-gmail)]

xrandr --prop
Screen 0: minimum 320 x 200, current 1680 x 1050, maximum 1680 x 1680
TMDS-1 connected 1680x1050+0+0 (normal left inverted right x axis y axis) 450mm x 280mm
        EDID_DATA:
                00ffffffffffff001ef98a0896410000
                350f0103802d1c78eaee95a3544c9926
                0f5054bfef0081806140b30001010101
                0101010101012d399030621a274068b0
                6e01c2181100001e000000fc00465044
                32313835570a20202020000000fd0038
                4c1f5110000a202020202020000000ff
                005435432035304e203136373930001e
   1680x1050 60.0*+ 60.0
   1280x1024 75.0 59.9
   1024x768 75.1 70.1 60.0 59.9
   832x624 74.6
   800x600 72.2 75.0 60.3 56.2
   640x480 75.0 72.8 66.7 60.0
   720x400 70.1

[Benjamin Pineau (ben-pineau)]

Other users of the G35 chipset reported the same problem on Ubuntu forums :
See http://ubuntuforums.org/showthread.php?t=647449&page=3

To sum up, with this G35 / X3500 based motherboard :

* HDMI connection to monitors works well
* VGA connection to monitors works well
* DVI connection to monitors works well under console and framebuffer
* DVI connection to monitors works well under Windows
* DVI connection to monitors works well under Xorg with the VESA driver
* DVI connection to monitors does not work under Xorg with the Intel driver

(By the way, if you're testing this hardware: that xorg bug seems also to
affect all G35 users : http://bugs.freedesktop.org/show_bug.cgi?id=11311 ).

[Benjamin Pineau (ben-pineau)]

Got it working!

I had the same problem (last time verified: saturday, 3 days ago), with the
same Asus P5E-VM HDMI motherboard. But today I re-tested "DVI out with intel
driver" and... it works, now! I didn't any change in xorg.conf, didn't
updated the kernel nor any xorg driver/server since saturday.

The only change I can think of: I upgraded the motherboard BIOS (from
revision 0202 to revision 0405). Could you test this too ?

Notes:
- I use Ubuntu Hardy (development version), wich has recent git versions of
  the drivers : linux kernel 2.6.24-rc6, xserver-xorg-video-intel version
  2.2.0+git20080107-1ubuntu1, xserver-xorg-core 1.4.1~git20080105-1ubuntu1,
  mesa 7.0.2-3ubuntu1, libxrandr2 1.2.2
- As stated above, when DVI wasn't working with intel xorg driver, VGA worked
  with intel driver, and DVI worked fine using the vesa xorg driver. Now
  everything work as expected.
- Attached xorg.conf, Xorg.0.log, and xrandr --verbose output using intel
  driver with DVI attached monitor.
- With DVI (but not with VGA), fonts are somehow blurry, and colors too pinky
  (maybe because my monitor is limited, but the "autoconfig" button does not
  work with dvi)
- I still need `Option "AccelMethod" "XAA"` else fonts and icons aren't
  displayed (same problem with VGA).
- In case it rings a bell: the HDMI+DVI adapter/transmitter chipset used in
  this G35/GMA X3500 based motherboard is a Silicon Image SILC SiI1392. See :
  http://www.siliconimage.com/products/product.aspx?id=106
  and the photo here :
  http://plusd.itmedia.co.jp/pcuser/articles/0711/30/news070.html
- DVI works, but the xorg logs reports a few errors related to SDVO :

ben$egrep '(EE|WW)' /var/log/Xorg.0.log | grep -v AIGLX
        (WW) warning, (EE) error, (NI) not implemented, (??) unknown.
(WW) The directory "/usr/share/fonts/X11/cyrillic" does not exist.
(II) Loading extension MIT-SCREEN-SAVER
(WW) intel: No matching Device section for instance (BusID PCI:0:2:1) found
(EE) intel(0): Unable to write to SDVOCTRL_E for SDVOB Slave 0x70.
(WW) intel(0): Register 0x70180 (DSPACNTR) changed from 0x14000000 to
0x58000000
(WW) intel(0): DSPACNTR before: disabled, pipe A
(WW) intel(0): DSPACNTR after: disabled, pipe A
(WW) intel(0): Failed to set up write-combining range (0xd0000000,0x10000000)
(EE) intel(0): Unable to write to SDVOCTRL_E for SDVOB Slave 0x70.
(EE) intel(0): Unable to write to SDVOCTRL_E for SDVOB Slave 0x70.
(EE) intel(0): Unable to write to SDVOCTRL_E for SDVOB Slave 0x70.
(EE) intel(0): Unable to write to SDVOCTRL_E for SDVOB Slave 0x70.
(EE) intel(0): Unable to write to SDVOCTRL_E for SDVOB Slave 0x70.

[Benjamin Pineau (ben-pineau)]

Created an attachment (id=13721)
Xorg.0.log with intel driver and working DVI connected monitor

[Benjamin Pineau (ben-pineau)]

Created an attachment (id=13722)
xorg.conf intel driver and working DVI out (used to get above logs)

[Benjamin Pineau (ben-pineau)]

Created an attachment (id=13723)
xrandr --verbose with intel driver and working DVI out

[Benjamin Pineau (ben-pineau)]

Created an attachment (id=13724)
lspci -vvvxx with working DVI out (and after upgrade to BIOS 0405)

[Benjamin Pineau (ben-pineau)]

Forgot to ask for this : Kevin,
if you use another BIOS version (and before you upgrade), could
you please run as root : "lspci -vvvxx" and compare/diff with
my version attached to this bug ? Thanks.

[Benjamin Pineau (ben-pineau)]

Too bad I didn't save a lspci before upgrading the BIOS.
Anyway, I saved a pre BIOS upgrade Xorg.0.log, and the
"PCI Scan" section has only one little difference :

before (bios 202, non working dvi):
(II) PCI: 00:1b:0: chip 8086,293e card 1043,8277 rev 02 class 04,03,00 hdr 00
after (bios 405, working dvi):
(II) PCI: 00:1b:0: chip 8086,293e card 1043,829f rev 02 class 04,03,00 hdr 00

So after bios update, "card 1043,8277" became "card 1043,829f" for this
device. Oddly, this 00:1b seems to be the audio controller:

ben$ lspci -s 00:1b
00:1b.0 Audio device: Intel Corporation 82801I (ICH9 Family) HD Audio
Controller (rev 02)

Note this is using HDMI-to-DVI, and that on this motherboard, the audio
chipset supports digital audio through HDMI.

Could the audio chipset configuration affect xorg DVI output ?
If so, could this problem find a workaround in kernel/xorg so
users of an older BIOS revision don't hit the bug ?

I attach a more detailed "sudo lspci -vvvxxxx" (with 4 "x" rather than 2,
and 3 "v" rather than 2), that would better show the differences (Kevin,
if you could attach the a "-vvvxxxx" rather than the previously asked
"-vvxx", that'd be nice, thanks).

[Benjamin Pineau (ben-pineau)]

Created an attachment (id=13725)
"sudo lspci -vvvxxxx" with new bios and working DVI

[Kevin DeKorte (kdekorte-gmail)]

Created an attachment (id=13729)
This is the lspci -vvvxx from a machine with a non-working HDMI

[Kevin DeKorte (kdekorte-gmail)]

I upgraded my BIOS to 405 like Ben and I still have issues.

[Kevin DeKorte (kdekorte-gmail)]

Created an attachment (id=13731)
This is after upgrading to BIOS 405

Note: if I try to disable HD Audio on my machine, the machine will not boot.

[Kevin DeKorte (kdekorte-gmail)]

Created an attachment (id=13732)
Xorg.0.log after BIOS upgrade from 301 to 405

Xorg.0.log after BIOS upgrade from 301 to 405

[Kevin DeKorte (kdekorte-gmail)]

intel driver appears to know that the DVI cable is connected and disconnected...

xrandr outputs with disconnected and connected

 xrandr
Screen 0: minimum 320 x 200, current 1680 x 1050, maximum 1680 x 1680
VGA connected 1680x1050+0+0 (normal left inverted right x axis y axis) 450mm x 280mm
   1680x1050 60.0*+ 60.0
   1280x1024 75.0 59.9
   1440x900 75.0 59.9
   1152x864 75.0 74.8 69.8 60.0
   1024x768 75.1 70.1 60.0
   832x624 74.6
   800x600 72.2 75.0 60.3 56.2
   640x480 75.0 72.8 66.7 60.0
   720x400 70.1
TMDS-1 disconnected 1680x1050+0+0 (normal left inverted right x axis y axis) 450mm x 280mm
[kdekorte@quad ~]$ xrandr
Screen 0: minimum 320 x 200, current 1680 x 1050, maximum 1680 x 1680
VGA connected 1680x1050+0+0 (normal left inverted right x axis y axis) 450mm x 280mm
   1680x1050 60.0*+ 60.0
   1280x1024 75.0 59.9
   1440x900 75.0 59.9
   1152x864 75.0 74.8 69.8 60.0
   1024x768 75.1 70.1 60.0
   832x624 74.6
   800x600 72.2 75.0 60.3 56.2
   640x480 75.0 72.8 66.7 60.0
   720x400 70.1
TMDS-1 connected 1680x1050+0+0 (normal left inverted right x axis y axis) 450mm x 280mm
   1680x1050 60.0*+ 60.0
   1280x1024 75.0 59.9
   1024x768 75.1 70.1 60.0 59.9
   832x624 74.6
   800x600 72.2 75.0 60.3 56.2
   640x480 75.0 72.8 66.7 60.0
   720x400 70.1

[Keith Packard (keithp)]

I suspect the BIOS is configuring the hardware differently now. We can fix the driver to do the same thing if we know what the differences are. If you add

Option "ModeDebug" "yes"

to the Device section of your xorg.conf file, it will dump lots of register information to the log file. Comparing broken vs non-broken machines may uncover something that we've neglected to configure in the driver.

[Kevin DeKorte (kdekorte-gmail)]

Created an attachment (id=13733)
With debug enabled

Log with ModeDebug enabled

[Kevin DeKorte (kdekorte-gmail)]

Keith, is there anything else I can provide to help solve this?

[Kevin DeKorte (kdekorte-gmail)]

Ben and I have been chatting and we're wondering if some of the problems could be due to Ben running in 32bit mode, while I'm running in 64bit mode. Could that be the reason why my hdmi interface does not work and his does?

[Benjamin Pineau (ben-pineau)]

Created an attachment (id=13776)
Xorg.0.log : working and connected DVI, with Option "ModeDebug" "yes"

[Benjamin Pineau (ben-pineau)]

Here is a Xorg.0.log with debug on my _working_ DVI out setup,
to compare with the Kevin's (broken DVI) one.

At the end of the day, the BIOS version seems unrelated to the pb.
Kevin (and someone else on the Ubuntu forums) upgraded to the same
version than me, but that didn't fixed the problem. So I don't know
what I changed to get the DVI out working. Very sorry.

Note that even if DVI is working here, the display with DVI isn't
very good : everything is blurry (esp. the fonts), and the colors
pinky (esp. the greys parts). Switching back to VGA (with no config
change), and everything displays properly and clearly again.

Another weird difference with Kevin : I need to explicitly force XAA
accel, else (with EXA) I get unusable display (no font and no icons
are rendered, for instance). On the opposite, Kevin told me XAA is
broken on his setup but EXA works well for him.

[Kevin DeKorte (kdekorte-gmail)]

I've been reading about this board on the Asus forums and it seems there are even problems with Vista and the HDMI->DVI convertor. So I decided, I'll take the HDMI->DVI convertor out of the equation. So I went to the local BestBuy and got a brand new HDMI->DVI cable.

Booted up the machine and the Bios screen works and the kernel messages work, until the intel driver is loaded. After that no signal.

Anything else I can provide?

[Kevin DeKorte (kdekorte-gmail)]

I was reading the asus forum for this bug and I found this post and response.

http://vip.asus.com/forum/view.aspx?id=20080122073058906&board_id=1&model=P5E-VM+HDMI&page=1&SLanguage=en-us

It points to a fix for the EDID parser in the Windows driver that is updated here

http://support.intel.com/support/graphics/sb/CS-028366.htm

Is this possibly related to this bug. I would assume that the EDID parser in the Linux driver has be checked for this problem.

[Michael Fu (michael-fu-intel)]

This bug has got a lot of information from our reporters. thanks!

I'm re-assign this bug to Hong.

[Kevin DeKorte (kdekorte-gmail)]

Created an attachment (id=14424)
intel_reg_dumper run in 1280x1024 mode

Run after reboot, with drivers from git and VGA and HDMI->DVI adapter plugged in

[Kevin DeKorte (kdekorte-gmail)]

Created an attachment (id=14425)
intel_reg_dumper run in 1280x1024 mode with vesa driver

Run after reboot with VGA and HDMI->DVI cables plugged in to same monitor

[Hong-liu (hong-liu)]

(In reply to comment #28)
> Created an attachment (id=14424) [details]
> intel_reg_dumper run in 1280x1024 mode
>
> Run after reboot, with drivers from git and VGA and HDMI->DVI adapter plugged
> in
>

SDVOB: 0x40080084 (disabled, pipe B, stall disabled, detected)

The SDVOB output is disabled when using intel driver while it is enabled in the vesa driver. It is weird since driver enabled it during startup.

What is the output of xrandr when SDVOB is disabled?
Or can you try to run intel_reg_dump frequently to see when the SDVOB is disabled?

in comment 18
> intel driver appears to know that the DVI cable is connected and
> disconnected...
So you are connect/disconnect the DVI cable and xrandr reports the correct states?

[Kevin DeKorte (kdekorte-gmail)]

>
> in comment 18
> > intel driver appears to know that the DVI cable is connected and
> > disconnected...
> So you are connect/disconnect the DVI cable and xrandr reports the correct
> states?
>

Yes, that is correct. I can plug in the HDMI/DVI cable and xrandr will show the connection as active and I can unplug it and it show the connection as disabled

[Kevin DeKorte (kdekorte-gmail)]

(In reply to comment #30)
> (In reply to comment #28)
> > Created an attachment (id=14424) [details] [details]
> > intel_reg_dumper run in 1280x1024 mode
> >
> > Run after reboot, with drivers from git and VGA and HDMI->DVI adapter plugged
> > in
> >
>
> SDVOB: 0x40080084 (disabled, pipe B, stall disabled, detected)
>
> The SDVOB output is disabled when using intel driver while it is enabled in the
> vesa driver. It is weird since driver enabled it during startup.
>
> What is the output of xrandr when SDVOB is disabled?

The output of xrandr when SDVOB is disabled and connected is that the display is plugged in and it does show the right monitor values. Just like in comment #1

> Or can you try to run intel_reg_dump frequently to see when the SDVOB is
> disabled?

How would you recommend doing this?

[Hong-liu (hong-liu)]

(In reply to comment #32)
> How would you recommend doing this?
>

The disable of SDVOB causes the no display issue. But it should be enabled by our driver, not sure why this is disabled.

Have you ever disabled the TMDS-1 output by xrandr?

Thanks,
Hong

[Kevin DeKorte (kdekorte-gmail)]

No I have not disabled TMDS-1. I have however ran the following commands and still not got a picture

xrandr --output TMDS-1 --auto

xrandr --output TMDS-1 --same-as VGA

[Kevin DeKorte (kdekorte-gmail)]

Created an attachment (id=14669)
This is same system, but with a ADD2 card installed

Things to note here. The TMDS-1 card is now the DVI connector and not the HDMI connector (where did the hdmi connector go?)

 xrandr --verbose
Screen 0: minimum 320 x 200, current 1680 x 1050, maximum 1920 x 1920
VGA disconnected (normal left inverted right x axis y axis)
 Identifier: 0x3b
 Timestamp: 33135
 Subpixel: unknown
 Clones:
 CRTCs: 0 1
TMDS-1 connected 1680x1050+0+0 (0x3d) normal (normal left inverted right x axis y axis) 450mm x 280mm
 Identifier: 0x3c
 Timestamp: 33135
 Subpixel: horizontal rgb
 Clones:
 CRTC: 0
 CRTCs: 0 1
 EDID_DATA:
  00ffffffffffff001ef98a0896410000
  350f0103802d1c78eaee95a3544c9926
  0f5054bfef0081806140b30001010101
  0101010101012d399030621a274068b0
  6e01c2181100001e000000fc00465044
  32313835570a20202020000000fd0038
  4c1f5110000a202020202020000000ff
  005435432035304e203136373930001e
  1680x1050 (0x3d) 146.4MHz +HSync +VSync
        h: width 1680 start 1784 end 1960 total 2240 skew 0 clock 65.3KHz
        v: height 1050 start 1056 end 1086 total 1089 clock 60.0Hz
  1920x1200 (0x3e) 154.0MHz +HSync -VSync
        h: width 1920 start 1968 end 2000 total 2080 skew 0 clock 74.0KHz
        v: height 1200 start 1203 end 1209 total 1235 clock 60.0Hz
  1920x1080 (0x3f) 138.5MHz +HSync -VSync
        h: width 1920 start 1968 end 2000 total 2080 skew 0 clock 66.6KHz
        v: height 1080 start 1083 end 1088 total 1111 clock 59.9Hz
  1680x1050 (0x40) 146.2MHz -HSync +VSync
        h: width 1680 start 1784 end 1960 total 2240 skew 0 clock 65.3KHz
        v: height 1050 start 1053 end 1059 total 1089 clock 60.0Hz
  1680x1050 (0x41) 119.0MHz +HSync -VSync
        h: width 1680 start 1728 end 1760 total 1840 skew 0 clock 64.7KHz
        v: height 1050 start 1053 end 1059 total 1080 clock 59.9Hz
  1600x1024 (0x42) 103.1MHz +HSync +VSync
        h: width 1600 start 1600 end 1656 total 1664 skew 0 clock 62.0KHz
        v: height 1024 start 1024 end 1029 total 1030 clock 60.2Hz
  1400x1050 (0x43) 155.8MHz +HSync +VSync
        h: width 1400 start 1464 end 1784 total 1912 skew 0 clock 81.5KHz
        v: height 1050 start 1052 end 1064 total 1090 clock 74.8Hz
  1400x1050 (0x44) 151.0MHz +HSync +VSync
        h: width 1400 start 1464 end 1656 total 1960 skew 0 clock 77.0KHz
        v: height 1050 start 1051 end 1054 total 1100 clock 70.0Hz
  1400x1050 (0x45) 145.1MHz -HSync +VSync
        h: width 1400 start 1496 end 1648 total 1896 skew 0 clock 76.5KHz
        v: height 1050 start 1051 end 1054 total 1093 clock 70.0Hz
  1400x1050 (0x46) 122.0MHz +HSync +VSync
        h: width 1400 start 1488 end 1640 total 1880 skew 0 clock 64.9KHz
        v: height 1050 start 1052 end 1064 total 1082 clock 60.0Hz
  1280x1024 (0x47) 135.0MHz +HSync +VSync
        h: width 1280 start 1296 end 1440 total 1688 skew 0 clock 80.0KHz
        v: height 1024 start 1025 end 1028 total 1066 clock 75.0Hz
  1280x1024 (0x48) 109.0MH...

[TheKorn (ubuntu-thekorn)]

Binary package hint: gdm

monitor: hanns-g hc194d
graphics chipset: intel G35
motherboard: asus p5e-vm hdmi
OS version: hardy alpha 5, patched on 3/5/2008

DISTRIB_ID=Ubuntu
DISTRIB_RELEASE=8.04
DISTRIB_CODENAME=hardy
DISTRIB_DESCRIPTION="Ubuntu hardy (development branch)"

gdm:
  Installed: 2.20.3-0ubuntu6
  Candidate: 2.20.3-0ubuntu6
  Version table:
 *** 2.20.3-0ubuntu6 0
        500 http://us.archive.ubuntu.com hardy/main Packages
        100 /var/lib/dpkg/status

When monitor is connected via VGA D-sub cable, everything works great! When monitor is connected via DVI cable (w/ DVI-HDMI adapter), monitor shows "unsupported display mode".

FWIW, this exact combo works GREAT under gutsy (even under HDMI).

more info: I'm rebooting in between swapping cables, and at no point in time are both cables connected to the machine. Sequence is to power down, swap VGA for DVI cable (or vice-versa), power up, see BIOS and then ubuntu splash screens, then when the login screen should come up (I hear the bongo sound, and can log in blind or administer via SSH) that's when I get unsupported display mode. So I'm not crashing at all, it's just not exactly working quite right.

[TheKorn (ubuntu-thekorn)]

Here is the xorg.0.log when the monitor is connected via the VGA cable.

[TheKorn (ubuntu-thekorn)]

Correction... I just tried the gutsy live CD, and I guess this combo did NOT work via HDMI then, only VGA. My mistake!

[Eric Anholt (eric-anholt)]

bf6d59ee98127ac6447848c4202c54d861a510d4 on master is likely to fix this bug

[KeithB (keith-brindley)]

Running a build taken Sunday night (GMT) but still no luck here even with those commit changes.

Anything useful I can collect? From what I can see, the TVs problem is the refresh rate.

As a side issue (may or may not be related), any attempts to change the mode with xrandr result in the command line coming back with no changes. (Running via remote desktop at this point - but when running on VGA,xrandr mode changes trigger a refresh of the remote desktop and redraw of the whole lot.)

[Otakar-trunecek (otakar-trunecek)]

I have the same motherboard and DVI output over HDMI->DVI converter works fine in X. I'm running current Arch Linux, xorg-server version 1.4.0.90, xf86-video-intel 2.2.1, motherboard bios version 301.

[Kevin DeKorte (kdekorte-gmail)]

(In reply to comment #38)
> I have the same motherboard and DVI output over HDMI->DVI converter works fine
> in X. I'm running current Arch Linux, xorg-server version 1.4.0.90,
> xf86-video-intel 2.2.1, motherboard bios version 301.
>

Can you please post your Xorg.0.log file and the output of 'xrandr --prop'

Thanks

[Otakar-trunecek (otakar-trunecek)]

Created an attachment (id=15120)
Xorg.0.log with working DVI

[Otakar-trunecek (otakar-trunecek)]

Created an attachment (id=15121)
output of 'xrandr --prop'

[Daniel J Blueman (danielblueman)]

With my G35 Asus P5E-VM running Ubuntu 8.04 pre-rel x86-64 with Intel driver 2.2.1, I am seeing two symptoms, using the HDMI port w/ HDMI->DVI adapter:

 - dual-link DVI cable to projector
  -> unable to lock, so blank screen

 - single-link DVI cable to Belinea 1600x1200 LCD
  -> consistent off-by-2 Y size

It seems entirely conceivable that my projector doesn't accept resolutions that aren't expected and untested via the DVI-D connection; I can readily see this happening on HDMI/DVI TVs which support only particular resolutions and possibly on particular digital monitors.

Observations which make me tie the bug to the Intel driver:
 - during BIOS init and before Linux takes over, we see:
  - 640x480

 - during Ubuntu's booting and VESA framebuffer use:
  - 640x480

 - in X, testing various resolutions, the monitor reports:
  - 1600x1198
  - 1280x1022
  - 1280x958
  - 1152x862
  - 1024x766
  - 800x598
  - 640x478

All output in /var/log/Xorg.0.log, xrandr etc, points the correct resolution being used, so it seems like a chip configuration/register setup issue.

Let me know of useful information I can get and contribute!

[Daniel J Blueman (danielblueman)]

There is a good chance of this being caused by a simple off-by-1 or 2 coding error; what debug do I need to dump the relevant register writes, or register state?

[KeithB (keith-brindley)]

(In reply to comment #42)

> - in X, testing various resolutions, the monitor reports:
> - 1600x1198
> - 1280x1022
> - 1280x958
> - 1152x862
> - 1024x766
> - 800x598
> - 640x478
>
> All output in /var/log/Xorg.0.log, xrandr etc, points the correct resolution
> being used, so it seems like a chip configuration/register setup issue.
>
> Let me know of useful information I can get and contribute!
>

I'm not convinced it's as simple as the -2 theory. Running the VESA driver on mine, the driver thinks it's running at 1024x768 but the TV reckons 1026x769 (complete with annoying bright purple line for those 2 extra pixels!). The Intel driver has reverted back to completely refusing to work.

Current setup:
 xorg server 1.3.0 (Gutsy standard)
 intel driver 1 week old git checkout (9th March)
 BIOS 405.

I'm hoping to get hold of a DVI compatible PC monitor tomorrow on loan and then plan on having a play with the monitor and TV on different BIOS revisions to see if there is any difference there (assuming the monitor is happy enough with slightly off resolutions!)

If anyone can think of other tests of information gathering that would be useful please let me know.

Could there be something in xorg server 1.4.0 that's helping it? I could be wrong but of the working reports I've seen they seem to generally (if not all) be running 1.4.0 or above.

[Michael Fu (michael-fu-intel)]

(In reply to comment #35)
> Created an attachment (id=14669) [details]
> This is same system, but with a ADD2 card installed
>
> Things to note here. The TMDS-1 card is now the DVI connector and not the HDMI
> connector (where did the hdmi connector go?)
>

Kevin, could you explain more about the ADD2 card? I assume it's a SDVO DVI card, right?

>This same DVI cable and monitor were working fine on an Asus MiniPC with an
>i915 chipset, until the machine died.

in your comment# 1, you mentioned a Asus MiniPC. It sounds to me it also has a HDMI port ( and you connect it to the same monitor using the same HDMI->DVI cable, right?) Do you have a log of it with "modedebug" turns on?

[Kevin DeKorte (kdekorte-gmail)]

(In reply to comment #45)
>
> Kevin, could you explain more about the ADD2 card? I assume it's a SDVO DVI
> card, right?

Yes the ADD2 card is an SDVO DVI card.

Got mine off ebay and it looks like this

http://cgi.ebay.com/Silicon-Image-Orion-ADD2-N-Dual-Pad-x16-PCIe-X8760_W0QQitemZ180225854325QQihZ008QQcategoryZ3762QQssPageNameZWDVWQQrdZ1QQcmdZViewItem

>
> >This same DVI cable and monitor were working fine on an Asus MiniPC with an
> >i915 chipset, until the machine died.
>
> in your comment# 1, you mentioned a Asus MiniPC. It sounds to me it also has a
> HDMI port ( and you connect it to the same monitor using the same HDMI->DVI
> cable, right?) Do you have a log of it with "modedebug" turns on?
>

No, the minipc had only a DVI ouput on it. I do not have that machine anymore so unfortunately I can't run any tests on it.

[Michael Fu (michael-fu-intel)]

*** Bug 14787 has been marked as a duplicate of this bug. ***

[Michael Fu (michael-fu-intel)]

(In reply to comment #46)
> (In reply to comment #45)
> >
> > Kevin, could you explain more about the ADD2 card? I assume it's a SDVO DVI
> > card, right?
>
> Yes the ADD2 card is an SDVO DVI card.
>
> Got mine off ebay and it looks like this
>
> http://cgi.ebay.com/Silicon-Image-Orion-ADD2-N-Dual-Pad-x16-PCIe-X8760_W0QQitemZ180225854325QQihZ008QQcategoryZ3762QQssPageNameZWDVWQQrdZ1QQcmdZViewItem
>
>
> >
> > >This same DVI cable and monitor were working fine on an Asus MiniPC with an
> > >i915 chipset, until the machine died.
> >
> > in your comment# 1, you mentioned a Asus MiniPC. It sounds to me it also has a
> > HDMI port ( and you connect it to the same monitor using the same HDMI->DVI
> > cable, right?) Do you have a log of it with "modedebug" turns on?
> >
>
> No, the minipc had only a DVI ouput on it. I do not have that machine anymore
> so unfortunately I can't run any tests on it.
>

Hmm..you said "using the same cable", then I assume you are using the HDMI->DVI cable . If the MiniPC doesn't have a HDMI port, how do you connect the HDMI->DVI cable to the machine?

[Kevin DeKorte (kdekorte-gmail)]

The cable I had was a dual link DVI cable. The Asus MB comes with a HDMI->DVI adapter. So it was the same cable, but with the adapter attached.

[Ep-elmerproductions (ep-elmerproductions)]

I also had trouble with the HDMI/DVI output on this motherboard in combination with a Dell 24" screen (2405FPW). When connecting via the HDMI-DVI converter included with the motherboard, and a DVI cable, there is no image on the DVI, and the monitor actually 'locks up' because it cannot handle the signal sent to it (you have to pull out the cable in order to change the input choice or get the on screen menu).

Reading about the SDVO cards I thought I'd give that a try as last resort. So I got one and plugged in in, and connecting the DVI cable directly to this card I have no problems running the panel at full resolution (and with a sharp image).

If there is any logging that I can add that may help to fix the initial issue, let me know. I'm not an expert, but I have played around the xorg.conf and even managed to compile/install a driver myself, so I should be ok with moderate care :) [Though I must also add that I had some problems reverting back to an packaged driver so I'm not 100% sure my system is sufficiently clean to give good results. Will upgrade with a clean system once Kubuntu Hardy is out, should be soon]

[Tomasz-potega (tomasz-potega)]

A similar problem with an Asus P5E-V (ATX variant of the P5E-VM) + Asus MW221U LCD display.

When connecting the monitor using the bundled HDMI/DVI converter, I get slightly blurry image, with the display reporting 1680x1049 (this time only one scanline less than expected, but the resolution is lower, too).

Environment: 32 bit Fedora 8 with all the updates applied, various git driver checkouts.

If there is more information that might be of any value (logs, dumps) - I'll grab it and post here.

[Hein-PietervanBraam (hp)]

I have the same problems described above with my Asus P5E-V HDMI motherboard. If there is anything I can test, please let me know. This machine isn't particularly useful for me if I can't get the second head to work, so I might as well do some debugging should that be required.

I am trying to get my VGA and HDMI->DVI attached LCD panels to be dual-screen, as of yet I have been unable to get any output on the second (HDMI->DVI) head as soon as X starts, the screen goed blank and will not display anything, including the OSD, which to me suggests a bad mode of sorts.

[Ep-elmerproductions (ep-elmerproductions)]

(In reply to comment #50)
I did a clean boot of the official Ubuntu 8.04 'desktop' release (live CD) and both with the SDVO card and without the result is a complete lockup of the PC (black screen, LCD recons the resolution is 640x480, tapping num/capslock does not do anything, cannot switch to Ctrl-Alt-F1, etc.). This is on amd64 platform. The lockup seems to occur right after you can see it 'starting CUPS' in textmode, not 100% sure whether this is driver related (no logs available from the live boot), although it seems to match the result I got when upgrading the xorg/intel packages on a 7.10 install to those included with Hardy...

With a clean reinstall of the official 7.10 release it works fine with the SDVO card as mentioned before using the included 'intel' driver (1920x1200, no blurred images).

[Allison Karlitskaya (desrt)]

See also the possibly-quite-related bug 15370.

[Jean-Baptiste Lallement (jibel)]

Thanks for reporting this bug and any supporting documentation. Since this bug has enough information provided for a developer to begin work, I'm going to mark it as confirmed and let them handle it from here. Thanks for taking the time to make Ubuntu better!

[Niqooo (niqooo)]

I alos had the same problems with my Asus P5E-V HDMI motherboard. But HDMI -> DVI output is now working with intel git versions of the drivers from today.
I think it was related to this bug : #15766

[Hong-liu (hong-liu)]

(In reply to comment #55)
> I alos had the same problems with my Asus P5E-V HDMI motherboard. But HDMI ->
> DVI output is now working with intel git versions of the drivers from today.
> I think it was related to this bug : #15766
>

Thanks for the finding, can anyone try the latest driver to see if HDMI->DVI is working?

Thanks,
Hong

[Allison Karlitskaya (desrt)]

G35 on ASUS P5E-VM here with a DELL 20" DVI monitor (hooked up via the HDMI->DVI dongle).

this fix nails it. thanks so much.

[Allison Karlitskaya (desrt)]

this bug has been fixed in git upstream.

see this patch: http://gitweb.freedesktop.org/?p=xorg/driver/xf86-video-intel.git;a=commitdiff;h=beb72ae5aa053303f5cc419e9c9d7c6db964f160;hp=c7fee208fd46e143965ea173984d284e1eec2a9b

possibly a good candidate for SRU since it appears to have closed quite a number of upstream bugs (all sorts of things related to using the second monitor port in various configurations)

[Allison Karlitskaya (desrt)]

i just rolled a package based on cherry-picking that patch from upstream. this is not enough to fix the problem.

there is something else between 2.2.1 and that revision that fixes something else...

one thing to note, though: before that patch, nothing works at all.

[Jean-Baptiste Lallement (jibel)]

Setting it back to "confirmed" then. Ryan, Can you comment the upstream report please ?

[Allison Karlitskaya (desrt)]

I've contacted Hong Liu to find out if he knows more.

[ajonat (ajonat)]

I can also confirm this. I have a dell inspiron with a 965GM intel card and HDMI does not work. I'm trying to use a 32" Samsung LCD TV and it says "mode incompatible" (it seems very much alike "unsupported display mode" that TheKorn configuration shows).

[Allison Karlitskaya (desrt)]

I have a reply to my email but I'm in Istanbul for the next two weeks and won't be able to try to build a package until I get back (problem affects only my computer at home).

[ajonat (ajonat)]

Hi Ryan,
do you happen to know if this bug is fixed in intrepid?

[Jeremy Lucier (jrlucier)]

I'm having similar problems on my Intel 946GZ computer with the Ibex RC upgraded from Hardy. I created a separate bug regarding the issue, but it sounds like this might be part of it. My bug: https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/289883

I was attempting to use my Asus 22" monitor with it via both HDMI and DVI. Both failed with error messages forcing usage of the vesa driver.

[Bryce Harrington (bryce)]

Hi Ryan,

Have you had time to re-test the bug? Since the fix you indicated earlier has gone upstream and we've released Intrepid with it, I'm wondering if it's still an issue.

[Bryce Harrington (bryce)]

[Assuming it's solved, since there's been no reply and it's marked fixed upstream.]

[TheKorn (ubuntu-thekorn)]

Amazingly, I still have this equipment after all this time. I've since upgraded to 10.10, and I no longer experience this issue so yeah, I'd call it fixed.