[i965] Bad resolution over HDMI in Jaunty

[Jaroslav Šmíd]

I have Intel X3100 video card with HDMI output. "Display" configuration tool in Ubuntu (tested 8.10 and 9.04 versions) allows me to set only 720x480 (480p) and once in a blue moon it allows 1920x540 (1080i). My TV supports 720p (1280x720). It is possible to set 720p resolution in Windows.

I tried to generate modeline using "gtf 1280 720 60" and adding that mode using xrandr. When I set it using xrandr nothing is shown on TV and xrandr does not show any error.

[lspci]
00:00.0 Host bridge [0600]: Intel Corporation Mobile PM965/GM965/GL960 Memory Controller Hub [8086:2a00] (rev 03)
 Subsystem: Dell Device [1028:0254]
00:02.0 VGA compatible controller [0300]: Intel Corporation Mobile GM965/GL960 Integrated Graphics Controller [8086:2a02] (rev 03)
 Subsystem: Dell Device [1028:0254]

[Geir Ove Myhr (gomyhr)]

Thank you for reporting this bug.

Could you please attach the following as separate attachments:
- /var/log/Xorg.0.log
- the output of `lspci -vvnn`
- /etc/X11/xorg.conf if it is changed from the default minimum configuration

And please don't confirm your own bugs. I know I skipped this bug while going through new bugs a couple of days ago since the status indicated that somebody already had looked at it and confirmed it.

[Jaroslav Šmíd (jardasmid-gmail)]

Ok, I'll do it on wednesday, I have school today and tomorrow till night and lots of homeworks to do.

[Jaroslav Šmíd (jardasmid-gmail)]

xorg.conf is the default one

[Jaroslav Šmíd (jardasmid-gmail)]

It seems no one care and this is not going to be fixed for final release. What a pitty, this is the only reason I still have to keep windows on my harddrive. And I can tell you that it is not working with version 2.7 of intel driver either (tested on Fedora 11 rawhide).

[Geir Ove Myhr (gomyhr)]

Jaroslav, it's not that nobody cares, but there are currently 258 open bugs only in the intel driver alone and only a handful of people to care about them. Therefore there are some hard priorities that have to be done. This close to the release of Jaunty that means that stability issues have priorities before anything else. Besides, since you report that it doesn't work in 2.7 either, it has to be fixed upstream before a patch can be pulled to ubuntu, and it is very unlikely that this can happen before Jaunty is out.

You obviously care about getting this fixed, and even though most others focus on other things, you can do a number of things to get this fixed later (hopefully early enough that it works with Karmic).
- Help triage and resolve bugs in ubuntu. This frees up resources for others to sooner look at you favourite bug.
- Test the current upstream development version. If you run ubuntu you can do this by using the xorg-edgers PPA (https://launchpad.net/~xorg-edgers), otherwise you can look at http://intellinuxgraphics.org/ for how to get the latest source code from git and compile it yourself.
- If the bug still exists in git master, report the bug upstream according to the instructions in
http://intellinuxgraphics.org/how_to_report_bug.html .
This may require some significant investment of your time, but after all, you are the one who cares the most about this and the one who knows this particular bug best. I (and probably others) would be happy to assist as long as it happens in a not so demanding way.

Finally, complaining about people not caring enough about you particular problem is usually not a good idea (not here nor at other projects). It just makes people sad and more likely to ignore you.

[Geir Ove Myhr (gomyhr)]

Jaroslav, with the release of Jaunty behind us and the Karmic LiveCD images working sufficiently well that it can be used for testing, this is a good time to see if we can get something done with this bug.

Would you mind testing Karmic (either LiveCD/LiveUSB or installed) and doing the following:
- Add the line
  Option "ModeDebug" "true"
to the "Device" section of /etc/X11/xorg.conf
- Restart X by logging out and in again
- run `apport-collect 340212`
- if the apport-collect fails, please attach at least /var/log/Xorg.0.log here

Could you also specify the model of the computer and monitor and how they are connected (via cable directly of via an adapter/spitter etc.)?

[Jaroslav Šmíd (jardasmid-gmail)]

Computer: Dell Studio 1535 laptop
Monitor: LCD TV Sony Bravia 20" (I'll write the exact model tomorrow), connected directly via HDMI cable v1.3 compatible
I'll download, test Karmic live and write the exact model of TV tomorrow and send requested info then.

[Jaroslav Šmíd (jardasmid-gmail)]

http://cdimage.ubuntulinux.org seems to be very slow for me (15kBps) now. It was always so fast, but now .. I cannot download the image. I will wait until it is faster, or if someone can link faster mirror (x86_64 version) or give me mirror list, I will download it from the mirror.

[Geir Ove Myhr (gomyhr)]

I have also been noticing the occasional slowdown. Yesterday, however, I was able to download alpha-2 at around 200 kBps. Since alpha-2 is relatively recent, you may use that (http://cdimage.ubuntu.com/releases/9.10/alpha-2/). Unlike the daily-live images, it also has a bittorrent feed, and that might be quicker.

[Jaroslav Šmíd (jardasmid-gmail)]

Torrent ... now I get 850kBps so I should be able to test it today.

[Jaroslav Šmíd (jardasmid-gmail)]

I cannot get wireless network working on LiveCD (first I connected to LAN, installed offered Broadcom B43 wireless driver, but after logout and even after restarting networking service, network manager would not show wireless networks) so that I can only attach Xorg.0.log.

[Jaroslav Šmíd (jardasmid-gmail)]

And the TV model is KDL-20S4000 (20 inch Sony Bravia, "Display" configuration tool shows it as 32inch)

[Jaroslav Šmíd (jardasmid-gmail)]

I see I did it in vain, nothing has changed even now with 2.8 driver available. Why do you tell me to download the LiveCD, test it, send that log and then do nothing? And you keep bug marked as incomplete just to have it expired and don't have to care about it...

[Geir Ove Myhr (gomyhr)]

I'm sorry about not following your updated information. I really meant to follow up on this when the updated information came, but other things kept me from paying attention to any bug reports the weeks following your updated logs. When I returned, I had lost the overview of what I was keeping an eye on before.

Note that when searching for bug reports in Launchpad, we distinguish between Incomplete bugs with and without answers. We often filter out incomplete bugs without reply, but an incomplete bug with a reply is one that needs attention (you will therefore see that sometimes we set an Incomplete bug to New and back to Incomplete again if there are replies that does not give the needed information). In an ideal world, another bug triager would have seen this bug was marked as Incomplete and had a reply and would have followed up in my absence. In practice, there are too few bug triagers for this to happen very often.

So what now? I'll have to go through the details of this again. I think it was ready for sending upstream after your last log, but a lot of things have changed in Karmic since then, so I may have to ask you to test an updated LiveCD, even though I understand that it may be frustrating.

[Jaroslav Šmíd (jardasmid-gmail)]

Architecture: amd64
DistroRelease: Ubuntu 9.10
LiveMediaBuild: Ubuntu 9.10 "Karmic Koala" - Alpha amd64 (20090801)
MachineType: Dell Inc. Studio 1535
NonfreeKernelModules: wl
Package: xserver-xorg-video-intel 2:2.8.0-0ubuntu1
PackageArchitecture: amd64
ProcCmdLine: BOOT_IMAGE=/casper/vmlinuz file=/cdrom/preseed/ubuntu.seed boot=casper initrd=/casper/initrd.lz quiet splash --
ProcEnviron:
 PATH=(custom, no user)
 LANG=en_US.UTF-8
 SHELL=/bin/bash
ProcVersionSignature: Ubuntu 2.6.31-4.23-generic
RelatedPackageVersions:
 xserver-xorg 1:7.4+3ubuntu5
 libgl1-mesa-glx 7.5-1ubuntu1
 libdrm2 2.4.12-1ubuntu1
 xserver-xorg-video-intel 2:2.8.0-0ubuntu1
 xserver-xorg-video-ati 1:6.12.99+git20090629.f39cafc5-0ubuntu4
Uname: Linux 2.6.31-4-generic x86_64
UserGroups:

XorgConf:
 Section "Device"
  Identifier "DefaultDevice"
  Option "ModeDebug" "true"
 EndSection
dmi.bios.date: 12/04/2008
dmi.bios.vendor: Dell Inc.
dmi.bios.version: A06
dmi.board.name: 0M263C
dmi.board.vendor: Dell Inc.
dmi.chassis.type: 8
dmi.chassis.vendor: Dell Inc.
dmi.modalias: dmi:bvnDellInc.:bvrA06:bd12/04/2008:svnDellInc.:pnStudio1535:pvr:rvnDellInc.:rn0M263C:rvr:cvnDellInc.:ct8:cvr:
dmi.product.name: Studio 1535
dmi.sys.vendor: Dell Inc.
fglrx: Not loaded
system:
 distro: Ubuntu
 architecture: x86_64kernel: 2.6.31-4-generic

[Geir Ove Myhr (gomyhr)]

Forwarding a bug report from ubuntu user Jaroslav Šmíd:
https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/340212

[Problem description]
An LCD TV Sony Bravia 20" (model: KDL-20S4000) connected to the HDMI port in dual-head setup with LVDS only gets a resolution of 720x480 (480p) instead of the supported resolution of 1280x720 (720p). This was originally reported using UMS but the attached log files are for KMS with kernel 2.6.31-rc4 and the newest 2.8.0 intel driver.

[Original description]

I have Intel X3100 video card with HDMI output. "Display" configuration tool in Ubuntu (tested 8.10 and 9.04 versions) allows me to set only 720x480 (480p) and once in a blue moon it allows 1920x540 (1080i). My TV supports 720p (1280x720). It is possible to set 720p resolution in Windows.

I tried to generate modeline using "gtf 1280 720 60" and adding that mode using xrandr. When I set it using xrandr nothing is shown on TV and xrandr does not show any error.

[lspci]
00:00.0 Host bridge [0600]: Intel Corporation Mobile PM965/GM965/GL960 Memory Controller Hub [8086:2a00] (rev 03)
 Subsystem: Dell Device [1028:0254]
00:02.0 VGA compatible controller [0300]: Intel Corporation Mobile GM965/GL960 Integrated Graphics Controller [8086:2a02] (rev 03)
 Subsystem: Dell Device [1028:0254]

[other system information]

Architecture: amd64
DistroRelease: Ubuntu 9.10
LiveMediaBuild: Ubuntu 9.10 "Karmic Koala" - Alpha amd64 (20090801)
MachineType: Dell Inc. Studio 1535
NonfreeKernelModules: wl
Package: xserver-xorg-video-intel 2:2.8.0-0ubuntu1
PackageArchitecture: amd64
ProcCmdLine: BOOT_IMAGE=/casper/vmlinuz file=/cdrom/preseed/ubuntu.seed boot=casper initrd=/casper/initrd.lz quiet splash --
ProcEnviron:
 PATH=(custom, no user)
 LANG=en_US.UTF-8
 SHELL=/bin/bash
ProcVersionSignature: Ubuntu 2.6.31-4.23-generic
RelatedPackageVersions:
 xserver-xorg 1:7.4+3ubuntu5
 libgl1-mesa-glx 7.5-1ubuntu1
 libdrm2 2.4.12-1ubuntu1
 xserver-xorg-video-intel 2:2.8.0-0ubuntu1
 xserver-xorg-video-ati 1:6.12.99+git20090629.f39cafc5-0ubuntu4
Uname: Linux 2.6.31-4-generic x86_64
UserGroups:

XorgConf:
 Section "Device"
  Identifier "DefaultDevice"
  Option "ModeDebug" "true"
 EndSection
dmi.bios.date: 12/04/2008
dmi.bios.vendor: Dell Inc.
dmi.bios.version: A06
dmi.board.name: 0M263C
dmi.board.vendor: Dell Inc.
dmi.chassis.type: 8
dmi.chassis.vendor: Dell Inc.
dmi.modalias: dmi:bvnDellInc.:bvrA06:bd12/04/2008:svnDellInc.:pnStudio1535:pvr:rvnDellInc.:rn0M263C:rvr:cvnDellInc.:ct8:cvr:
dmi.product.name: Studio 1535
dmi.sys.vendor: Dell Inc.
fglrx: Not loaded
system:
 distro: Ubuntu
 architecture: x86_64kernel: 2.6.31-4-generic

[Attachments downstream]
http://launchpadlibrarian.net/29793890/XorgLog.txt
http://launchpadlibrarian.net/29793892/Xrandr.txt
http://launchpadlibrarian.net/28020986/Xorg.0.log (log with UMS and slightly older driver version)
There's dmesg output and a few other possibly relevant logs in the downstream report.

[Possibly irrelevant information]
With KMS the TV on the HDMI port is reported as DVI instead of HDMI. In this case only the first block of EDID information is registered. The log with UMS shows the full EDID information.

[Geir Ove Myhr (gomyhr)]

I have reported the bug upstream: http://bugs.freedesktop.org/show_bug.cgi?id=23098
Please register at freedesktop.org and add yourself in the CC: field so that you can reply to any requests from the intel developers.

[Geir Ove Myhr (gomyhr)]

I came to think of a few pieces of additional information that might be useful for the upstream developers. They might ask for it, but you may also upload it right away:
1. Kernel Mode Setting is now used in Karmic, so it is actually a kernel module that takes care of selecting the resolution. This code has been ported from the intel driver. It is possible to switch back to Userland Mode Setting (UMS) by adding this the boot parameter i915.modset=0 when booting. The code in the intel driver will then be used to set the resolution. This should select the same resolution as with KMS, but unlike the kernel module the intel driver dumps extra information to Xorg.0.log when the ModeDebug line is present is xorg.conf (as you have). So if you could boot with UMS and attach the resulting Xorg.0.log and the output of `xrandr --verbose` that may be valuable information.
2. You report that this happens when you use both the internal laptop monitor (LVDS) and the TV as an external monitor. Does it also happen if you disable the LVDS? You can do that by adding the following section to xorg.conf:
Section "Monitor"
      Identifier "My internal monitor"
      Option "Ignore" "True"
EndSection
and the following line to the "Device" section:
     Option "monitor-LVDS" "My internal monitor"
(with KMS it may be "monitor-LVDS1", but I'm not sure...)

[Jaroslav Šmíd (jardasmid-gmail)]

Created an attachment (id=28299)
Xorg log with KMS disabled

[Jaroslav Šmíd (jardasmid-gmail)]

Created an attachment (id=28300)
"xrandr --verbose" output with KMS disabled

[Jaroslav Šmíd (jardasmid-gmail)]

Well, that didn't go too well. Got system freeze when the internal display was disabled, nothing was shown on TV and couldn't access logs. I will try to run X from the command line and redirect stdout and stderr to flash drive and hope it gets synchronized before hard reboot.

[Geir Ove Myhr (gomyhr)]

Do you get the system freeze both with KMS and UMS? The reason I think it is useful information to try with the LVDS display disabled is to see if the low resolution on the TV is because of the dual-head configuration or not. Another option would be to connect the TV to a computer with a non-mobile chipset (like 965G) if you have that around, to see if it is an issue there. Actually, if you have any other computers around with intel chipsets and HDMI output it could be useful data points to know whether the problem occur when the TV is connected there.

Btw, it is a known issue (i.e. listed on http://intellinuxgraphics.org/2009Q2.html) that with KMS interlaced modes are not supported. That's the reason there is one more mode on the DVI/TMDS in your xrandr output when using UMS. So with KMS you won't be able to use 1080i at all until it is implemented.

[Jaroslav Šmíd (jardasmid-gmail)]

Freeze was with UMS and I'm going to get mad when I boot that LiveCD again and test it with KMS :-) ... Ok, I will, but perhaps tomorrow.

[Geir Ove Myhr (gomyhr)]

I thought that if it was with KMS, it was a chance it would not crash with UMS. The other way, I have less hope. If your sanity is in jeopardy, I suggest you skip it (at least until upstream asks for it).

Of course we would also appreciate the freeze problem being troubleshot and reported, but not in this bug report.

[Michael Fu (michael-fu-intel)]

Created an attachment (id=28391)
xorg.log with KMS enabled

[Michael Fu (michael-fu-intel)]

1) to get 1080i mode used by X, we need to support interlace mode (dup of bug# 11220)
2) to read out 720p mode , we need to support to parse mode from EDID extensions for HDMI.

[Geir Ove Myhr (gomyhr)]

(In reply to comment #4)
> 1) to get 1080i mode used by X, we need to support interlace mode (dup of bug#
> 11220)

Yes, I noted that in the downstream report. Let this bug focus on 720p.

[Michael Fu (michael-fu-intel)]

pls try if this patch work or not, for 720p problem:

http://<email address hidden>/msg42703.html

[Jaroslav Šmíd (jardasmid-gmail)]

(In reply to comment #6)
> pls try if this patch work or not, for 720p problem:
>
> http://<email address hidden>/msg42703.html
>

I compiled kernel (2.6.29) with the patch applied (I had to make some some changes - remove "timing_level = standard_timing_level(edid);" and remove "timing_level" argument from "drm_mode_std(dev, std, timing_level);" because it didn't compile with that. But ... it's still the same, only 480p available.

[Jaroslav Šmíd (jardasmid-gmail)]

I was asked (https://bugs.freedesktop.org/show_bug.cgi?id=23098) to try kernel patch (http://<email address hidden>/msg42703.html) if it solves the problem. Is it possible to include it in ubuntu's daily live (at least temporarly) so that I can test it?

[Jaroslav Šmíd (jardasmid-gmail)]

Forget it, compiled custom kernel, doesn't work.

[Jaroslav Šmíd (jardasmid-gmail)]

(In reply to comment #7)
> (In reply to comment #6)
> > pls try if this patch work or not, for 720p problem:
> >
> > http://<email address hidden>/msg42703.html
> >
>
> I compiled kernel (2.6.29) with the patch applied (I had to make some some
> changes - remove "timing_level = standard_timing_level(edid);" and remove
> "timing_level" argument from "drm_mode_std(dev, std, timing_level);" because it
> didn't compile with that. But ... it's still the same, only 480p available.
>

I'll try to compile 2.6.31 from git to see if it works, but I doubt it.

[Jaroslav Šmíd (jardasmid-gmail)]

Grr, I don't get it. Against which version of kernel is that patch? I downloaded 2.6.31rc8 and there are only 2 arguments to drm_mode_std as in 2.6.29.

[Jaroslav Šmíd (jardasmid-gmail)]

(In reply to comment #9)
> Grr, I don't get it. Against which version of kernel is that patch? I
> downloaded 2.6.31rc8 and there are only 2 arguments to drm_mode_std as in
> 2.6.29.
>

Ok, found that. I applied http://<email address hidden>/msg42703.html on top of patch provided by yakui zhao adding third argument for drm_mode_std (http://patchwork.kernel.org/patch/36729/mbox/) and two patches adding drm_cvt_mode (http://patchwork.kernel.org/patch/36727/mbox/) with fix (http://<email address hidden>/msg41991.html) and drm_gtf_mode (http://patchwork.kernel.org/patch/36728/mbox/). Compiled 2.6.31rc8 with them applied and what I get now is WORKING 720p resolution. 720p is available with KMS only, it doesn't appear when using UMS, but I don't need UMS (but I can test patch adding support for it if someone create one).

Thank you a lot, really. Finally working 720p :-)

[Jaroslav Šmíd (jardasmid-gmail)]

Created an attachment (id=29026)
xrandr --verbose output with KMS enabled showing 720p resolution available

[Jaroslav Šmíd (jardasmid-gmail)]

It does now, tried with kernel-2.6.31rc8 and it works with KMS. See the upstream bugreport. It would be great if those kernel patches were included in Ubuntu 9.10.

[Michael Fu (michael-fu-intel)]

(In reply to comment #10)
 720p
> is available with KMS only, it doesn't appear when using UMS, but I don't need
> UMS (but I can test patch adding support for it if someone create one).
>

UMS need fixes in xserver rather than just xfree86-video-intel. we've patches but no response and we've given up...

[Michael Fu (michael-fu-intel)]

btw, thanks for the testing and I'll reopen this bug till the patch falls into Dave's tree..

[Geir Ove Myhr (gomyhr)]

Thank you for building and testing, Jaroslav.

It is usually the developers that mark the upstream bugs as Fix Released, and usually only when the patches have been committed to a source code repository. Once we see where the fix is commited to, we can see how feasable it is to pull it into Karmic. As you noticed it doesn't apply directly to -rc8.

[Geir Ove Myhr (gomyhr)]

I see from the upstream comment by Michael Fu that it is aimed for Dave Airlie's tree. IIRC this is for the next kernel cycle, which means that it will only hit the mainline kernel for 2.6.32. Again, someone else has to decide how feasible it will be to pull this into Karmic.

[Jaroslav Šmíd (jardasmid-gmail)]

This is patch for 2.6.31-rc8 composed of all those patches mentioned in upstream bug. I see no reason not tu pull it to Karmic even for 2.6.30 kernel. But this is perhaps normal. When there is a patch fixing some problem, one must wait 1year for it to be included in "his" distribution. As I switched to Fedora, I will ask there, maybe they will be willing to include it in F12.
Thanks anyway.

[yakuizhao (yakui-zhao)]

The following commit is already put in the Dave's drm-next tree.
commit 882f0219518196a94cd2772004e87b178467139a
Author: Zhao Yakui <email address hidden>
Date: Wed Aug 26 18:20:49 2009 +0800

    drm/kms: Parse the detailed time info in CEA-EDID

So this bug can be marked as resolved.
thanks.

[Michael Fu (michael-fu-intel)]

*** Bug 13025 has been marked as a duplicate of this bug. ***

[Bryce Harrington (bryce)]

This was marked upstream with a kernel patch that has gone into the drm-next tree. I assume it is already incorprated into Karmic's kernel by now but will leave this for the kernel team to verify.

[Leann Ogasawara (leannogasawara)]

Seems the patch to resolve this went in post 2.6.31. As an interim solution you can try using one of the mainline kernel builds which should contain this patch - For ex http://kernel.ubuntu.com/~kernel-ppa/mainline/v2.6.32-rc4/

ogasawara@yoji:~/linux-2.6$ git show 882f0219518196a94cd2772004e87b178467139a

commit 882f0219518196a94cd2772004e87b178467139a

Author: Zhao Yakui <email address hidden>

Date: Wed Aug 26 18:20:49 2009 +0800

    drm/kms: Parse the detailed time info in CEA-EDID

[Guido Günther (agx)]

I can confirm this is working now nicely with 2.6.32 and kms enabled.