[i965gm] lower reslutions no longer available on LVDS with 2.7.99 (KMS)

[Ludovico Cavedon]

Binary package hint: xserver-xorg-video-intel

Since the upgrade to the new intel xorg driver 2.7.99 from xorg-edgers on janty and the enabling of kernel modesetting, lower resolution for my laptop screen are no longer available:

Current output:
$ xrandr
Screen 0: minimum 320 x 200, current 1280 x 800, maximum 8192 x 8192
VGA1 disconnected (normal left inverted right x axis y axis)
LVDS1 connected 1280x800+0+0 (normal left inverted right x axis y axis) 304mm x 190mm
   1280x800 59.9*+
TV1 disconnected (normal left inverted right x axis y axis)

Before the upgrade also resolutions "1024x768", "800x600" and "640x480" were available. That means I can no longer mirror the LVDS with an external monitor with lower resolution (like a projector).

ProblemType: Bug
Architecture: amd64
DistroRelease: Ubuntu 9.04
Package: xserver-xorg-video-intel 2:2.7.99.901+git20090610.9d3c3b05-0ubuntu0sarvatt~jaunty
ProcEnviron:
 LANG=en_US.UTF-8
 SHELL=/bin/bash
ProcVersion: Linux version 2.6.30-020630-generic (root@zinc) (gcc version 4.2.3 (Ubuntu 4.2.3-2ubuntu7)) #020630 SMP Wed Jun 10 09:04:38 UTC 2009
SourcePackage: xserver-xorg-video-intel
Uname: Linux 2.6.30-020630-generic x86_64
UnreportableReason: This is not a genuine Ubuntu package

[lspci]
00:00.0 Host bridge [0600]: Intel Corporation Mobile PM965/GM965/GL960 Memory Controller Hub [8086:2a00] (rev 0c)
 Subsystem: Hewlett-Packard Company Device [103c:30cd]
00:02.0 VGA compatible controller [0300]: Intel Corporation Mobile GM965/GL960 Integrated Graphics Controller [8086:2a02] (rev 0c)
 Subsystem: Hewlett-Packard Company Device [103c:30cd]

[Geir Ove Myhr (gomyhr)]

From the attached Xrandr.txt it doesn't look like the EDID information has been successfully captured. Would you mind adding
  Option "ModeDebug" "true"
to the "Device" section of your xorg.conf and upload the resulting Xorg.0.log. You may also remove/comment out the other Option lines in that section, as they should all be redundant or meaningless with the new driver. Also, the file ~/.config/monitors.xml.txt might interfere with what the driver normally does, so for testing purposes, it would be nice if you can rename that. Finally, it would be nice if you could comment out the SubSection "Display" in your "Screen" section, just to eliminate that as a source of the problems (it may be necessary for dual-head, but let's simplify as much as possible for testing).

[Ludovico Cavedon (cavedon)]

I did all the changes requested. Here is the new Xorg.0.log

Thanks for the help!

[Ludovico Cavedon (cavedon)]

If I disable the kernel mode setting (i.e. remove the "modeset=1" parameter to the i915 kernel module), things go back to work properly.
Here is the xorg.0.log with kernel mode setting disabled.

[Geir Ove Myhr (gomyhr)]

Thank you for the additional information. From the logs its evident that my theory of EDID not being available was wrong. This bug looks more or less ready to send upstream. I'm not sure whether the missing modes is a bug or simply a missing feature that is well known. Therefore, I will mark the bug as ready for sending upstream for now, so that possibly someone with more knowledge about KMS than me can take a look at it before sending it upstream.

In the meantime there are a few details that may be useful for an upstream report that you could fill in:
- The make and model of you computer
- The output of `xrandr --verbose` when running with and without KMS (as attachments)
- the xorg.conf you are using now (as attachment). I know you described how you changed it, but it is nice to have it available as a file for upstream to look at.

[Ludovico Cavedon (cavedon)]

Geir Ove Myhr wrote:
> - The make and model of you computer

Manufacturer: Hewlett-Packard
Product Name: HP Pavilion dv2500 Notebook PC (dv2570el)

Thanks for the help!

[Geir Ove Myhr (gomyhr)]

Thank you for the extra information. It seems that KMS changes a few things and it's hard for me to say what are bugs and not. For example, with KMS the EDID information is no longer visible in the output of `xrandr --verbose` (but it is in Xorg.0.log) and the "ModeDebug" option does not produce as verbose a Xorg.0.log as without KMS. I guess it will take some time for triagers to get used to what is erratic behavior of KMS in the logs and what is normal.

I leave this as ready for upstreaming for now.

[Matt Zimmerman (mdz)]

I'm seeing similar behaviour here. I use my laptop with an external display, and usually use the external display in its native mode (1920x1200) and disable the internal LCD. With the latest Karmic bits, 1920x1200 is no longer available, and xrandr only shows the VESA modes despite the driver having captured correct EDID (as shown in Xorg.log):

VGA1 connected 1024x768+0+0 (normal left inverted right x axis y axis) 550mm x 340mm
   1024x768 60.0*
   800x600 60.3 56.2
   640x480 60.0
LVDS1 connected 1680x1050+0+0 (normal left inverted right x axis y axis) 331mm x 207mm
   1680x1050 60.0* 50.0

[Matt Zimmerman (mdz)]

I backed out to kernel 2.6.30, and this problem went away, so it seems to be a kernel regression. Marking as such.

[Geir Ove Myhr (gomyhr)]

Matt, are you sure you have the same problem as Ludovico? He reported this with:
ProcVersion: Linux version 2.6.30-020630-generic

[Matt Zimmerman (mdz)]

On Wed, Jul 08, 2009 at 12:40:31AM -0000, Geir Ove Myhr wrote:
> Matt, are you sure you have the same problem as Ludovico? He reported this with:
> ProcVersion: Linux version 2.6.30-020630-generic

No, I'm not certain. Thanks for pointing out that difference.

--
 - mdz

[Ludovico Cavedon (cavedon)]

Matt Zimmerman wrote:
> On Wed, Jul 08, 2009 at 12:40:31AM -0000, Geir Ove Myhr wrote:
>> Matt, are you sure you have the same problem as Ludovico? He reported this with:
>> ProcVersion: Linux version 2.6.30-020630-generic
>
> No, I'm not certain. Thanks for pointing out that difference.

It looks like a different problem.
My problem is that when KMS is enable *only* 1920x1200 is shown and VESA
modes are no longer shown, with kernel 2.6.30.

[Geir Ove Myhr (gomyhr)]

Ludovico, I was hoping that someone with more in-depth knowledge would come around and send this upstream, doing some quality control at the same time. Since that hasn't happened yet, I guess we should just go for it. Before sending it upstream it would be nice if you could test with the newest kernel included in karmic (2.6.31-1-generic) and the driver from x-updates (2:2.7.99.901+git20090702.74227141-0ubuntu1 available at https://launchpad.net/~ubuntu-x-swat/+archive/x-updates/). If you could upload the same information as before with these upgrades (xrandr --verbose, Xorg.0.log with ModeDebug, both with and without KMS), we can use those files for the upstream report (they like logs and test with recent versions).

[Robert Hooker (sarvatt)]

It should be fixed by the intel driver in x-updates.

http://cgit.freedesktop.org/xorg/driver/xf86-video-intel/commit/?id=246cec965958e94babf5377e6f221522b05fb458

[Robert Hooker (sarvatt)]

I should clarify that it is for karmic that the x-updates driver contains the fix, it has not been brought back to 2.7 branch of the intel driver to make a package for jaunty on there.

[Bryce Harrington (bryce)]

Fix is included in git20090702, which will be uploaded to karmic shortly.

[Launchpad Janitor (janitor)]

This bug was fixed in the package xserver-xorg-video-intel - 2:2.7.99.901+git20090702.74227141-0ubuntu1

---------------
xserver-xorg-video-intel (2:2.7.99.901+git20090702.74227141-0ubuntu1) karmic; urgency=low

  [Robert Hooker]
  * Update to git 20090702 (master branch) up to commit 74227141
   - Treat disabled CRTCs as "not covering" for scanline wait purposes.
     Fixes problems returning from a dpms off state.
     (LP: #390917, #383973, #390633, #385448, #388357)
   - Add an xorg.conf option to control swapbuffers behavior.
   - Set hot plug interrupt to detect HDMI output.
   - Add a pipe-a quirk for Dell mini
     (LP: #395306)
   - Add a pipe-a quirk for thinkpad x30
     (LP: #304614)
   - Reenable XvMC support in UXA. Now includes vld support.
   - Fix EDID for LVDS output device to add the default modes in UMS.
     (LP: #385832)
   - Export EDID when using KMS, so Display options shows info correctly
     (LP: #395140)
   - Fixes for SDVO LVDS mode detection.
   - Fix major performance regression of trapezoid rendering in UXA.
   - Fixes for hangs when GL compositing is used with multiple monitors
     with different refresh rates.
   - Support for new IGDNG devices.
   - Only get the VBIOS in non-KMS mode
   - Add a few error messages for DRM initialization
   - Don't try to pin buffers in KMS mode
   - Fix 945GM VT switch in UMS
   - Clear the bo on the rotate scratch pixmap (Fixes problems with rotation)
   - uxa: Fix segfault on source-only picture usage with FallbackDebug.
   - Various XvMC fixes.
  * Merge with Debian experimental. Remaining Ubuntu changes:
   - Add lpia architecture
  * Drop patches since they're already upstream:
    - 110_quirk_hp_mini.patch
    - 117_quirk_thinkpad_x30.patch
    - 118_pixmap_inline_funcs.patch
    - 119_fix_drawable_abuse.patch
    - 120_fix_dri2_vblank_syncing_segfault.patch
    - 121_dont_change_blank_sync_width.patch
    - 122_wait_for_vblank_to_sub_border_color.patch
    - 123_kms_export_edid.patch

 -- Bryce Harrington <email address hidden> Thu, 09 Jul 2009 14:02:45 -0700

[Steve Beattie (sbeattie)]

Marking the kernel task of this bug as closed, since it was apparently an issue with intel xorg driver (and fixed there as well). Please re-open if driver update did not fix the issue. Thanks for helping to improve Ubuntu!

[Ludovico Cavedon (cavedon)]

Well, yes, the bugs seems to be fixed, now I see all the VESA modes in the xrandr output.
But now if I try to enable the VGA output I get a black screen with just the mouse pointer... so not much useful...
However, this is another new bug. Thanks for the help!